IRS drift - Details please ?

[Pepperseed]

Hello there,

I'm studying IRS at the moment and would like to submit a question regarding what exactly is drifting during their operation.
My understanding is that full alignment consists in "calculating" attitude based on gravity sensing, calculating latitude based on earth rate (which is compared to pilots "entered" latitude), finding true north by comparing A/C axis to the direction "vertical to earth rotation axis".

Now, after a flight, drift will have occurred and will be sort of "measured" in terms of residual GS and position discrepancy which both will have to be within "acceptable" margins.

1) May I then ask what would be the manufacturer margins for a 737 classic (no GPS)

2) In the situation where a quick align is decided for the following sector - which I understand consists in merely "zeroing" position and residual GS, what exactly in the IRU is "left" with a drift subsequent to the previous sector ? Are these both attitude and True North orientation ?

3) 737 classic (no GPS):
What exactly does the IRS transfer switch when switched to one side with regards to the opposite side connection with the FMC ?
Let's say transfer switch is selected to "both left", what is resulting of this sole action with regards to the R IRS connection with the FMC ?

Many thanks

[underfire]

There are a lot of questions here.

The IRS will drift, unless it is updated. There are many ways that this can occur. Since it appears your focus is on non-GPS ac, it must be updated manually.
For procedure designs, (FAA) the IRU drift rates are assumed to be 8nm/hour, but this far exceeds other issues.

While powered up, the IRS will continue to calc drift until updated, this includes the taxi time/speed, and ground time (as the IRS will continue to calc location based on the rotation of the Earth, ie you are sitting on the ground for an hour and the position difference) until an update is entered.

Location is always entered true.

As far as absolute system tolerances, this is an example from Honeywell.

[A Squared]

Mybe this will help with the question of "what exactly is drifting" imagine a single axis "IRS" ie: it's just a single accelerometer on little cart which travels along a straight level rail. It calculates where the cart has moved to along the rail by measuring the cart's acceleration, and integrating that with time to calculate position along the rail. You move the cart from the zero point to the 10 meter point along the track. The accelerometer measures the cart's acceleration as you move it, and the IRS' computer calculates the cart's velocity, and from the velocity, computes the cart's position. If the accelerometer measures acceleration perfectly, and the computer computes perfectly, then the IRS should tell you that the cart's position is 10.00000000 meters, and it's velocity is 0.00000000 m/s. But, nothing is perfect, so there are slight errors in the accelerometer measurements, and slight errors in the computation, so when your cart is stopped at the 10.000 meter mark, the "IRS" says the cart is at the 10.000001 meter position and it's velocity is 0.000001 m/s. That, in a nutshell is "drift" your calculated position and velocity 'drifts" from true position and velocity. It;s the result of errors in the acceleration measurement and errors in the computational methods. In a real IRS there's many, many more sources of errors. You have 3 accelerometers, all with measurement errors, their axes are not perfectly aligned, you have gyros measuring rotation, imperfectly, and all these measurements are used to compute your vleocity and position in 3 dimensions, which inevitability is done imperfectly. so like the cart, only in 3 dimensions, your computed position drifts away from your true position.

[Pepperseed]

Many Thanks for the data and explanations.

Regarding the quick align thing, it sounds a bit odd to me that attitude and orientation (as I understand it's the case - I may be wrong) are allowed not to be reset at any possible occasion given their critical importance.

As for the transfer switch on the 73 CL, more precisely, my question is to know whether when selected to one side (out of "normal" mode), the opposite side IRS is still taken into account by the FMC for navigation if it does not show any fault...
Many Thanks

[underfire]

welcome to reality. Enjoy.

[Bergerie1]

Pepperseed,

This may help. Read the section on alignment:-
https://www.skybrary.aero/index.php/...on_System_(INS)

[FlyingStone]

The IRS transfer switch affects only the flight instruments (providing different source of information to EFIS symbol generator), not the FMC.

Boeing's recommendation is to always perform a full IRS align before each flight if time permits, and makes a lot of sense to do it with the GPS inop / not installed.

[Pepperseed]

Tks Bergerie
It may be some glitch on my side but the link shows an empty page....

Hi FlyingStone
Thanks for your accurate answer.
Would you then agree that the only ways for a 73 CL IRS to be removed from the NAV loop would be either by lack of elec power or by the system itself if the FMC Kalman filter detects an unacceptable drift from said IRS ?

[Alex Whittingham]

I'd be surprised if IRS position was updated in flight. In general the IRS position is an input to the FMS (together with GPS (or not), DME, and VOR in some cases). The FMS determines its position taking all data and known errors into account using the ubiquitous Kalman filter but doesn't at any point change or update the raw IRS position.

The one inertial system that I am aware of that could be position updated in flight was the Carousel INS. I remember the start of the technique was to select position and hold over a known position like a VOR overhead, memory fades beyond that point.

[Bergerie1]

Try this and then enter INS into the search space top right:-

https://www.skybrary.aero/index.php/Main_Page

[wiedehopf]

https://www.skybrary.aero/index.php/...on_System_(INS)

here we go
@bergerie1 there is a small button with a globe and a chainlink to insert stubborn links

[Pepperseed]

Hello Alex Whittingham

It is also my understanding that an IRS is not NAV updatable in flight. After the POS parking alignment, it "only" delivers its calculations to the FMC in a continuous manner without being updated. It is the FMC - downstream the IRS' - that refines the navigation by using other external positions sources that it processes through a mathematical model (Kalman Filter) to compute what is supposed to be the most likely position with some degree of trust showed by the ANP (actual navigation performance).
My question is more about on what exact criteria an IRS may be deemed unreliable by the FMC, and what is the order of size of the drift magnitude that would send a "fault" Master Caution to the Mide Selector Unit...
Cheers

[Goldenrivett]

Hi Alex Whittingham,

The one inertial system that I am aware of that could be position updated in flight was the Carousel INS

You never updated the INS raw position doing that, you simply applied a position bias. (As FMS does continuously)
At the end of the flight, we'd "flush the fix" to reveal the raw INS position again and compare the position error with the gate position.

[Pepperseed]

Thanks Guys for the Skybrary link.
Lots of good reading to do

[FlyingStone]

Originally Posted by Pepperseed

Would you then agree that the only ways for a 73 CL IRS to be removed from the NAV loop would be either by lack of elec power or by the system itself if the FMC Kalman filter detects an unacceptable drift from said IRS ?

Probably, yes. However, the FMC should disregard the IRS which is drifting away automatically. In any case, you should get the VERIFY POS: IRS-FMC, and then during the FMC Nav Check supplementary procedure, you can force the FMC to take one of the available sources as the current position (IRS L/R, RADIO), when you have confirmed that that is the most accurate source.



Note that as in the Honeywell document, for this chart to be used accurately, full IRS align is required before each flight.

[Pepperseed]

Thanks FlyingStone
Good stuff
Not rated on the 73 yet. Just starting to learn from documentation I can find here and there...
Is the VERIFY POS a message on the CDU ?
May I ask you a quick description of the FMC NAV check Supplementary procedure and where to find it in the Boeing Doc ? I suppose it's all carried out through the CDU ?

[30/30 Green Light]

Quote, "The one inertial system that I am aware of that could be position updated in flight was the Carousel INS. I remember the start of the technique was to select position and hold over a known position like a VOR overhead, memory fades beyond that point." Alex, I believe it was the "DME Update" We used it Oceanic over the Pacific back in the early 80's. Fly to an en route VOR/DME (Marshall Islands IIRC in our case) and press DME Update when overhead,(holding was not required). Max allowable cumulative displacement error at trip end was 3 + 3T NM, where T = Time in Nav mode. Cheers.

[mustafagander]

On the Carousel INS, DME updating required you to enter the DME update area of the INS and enter the lat/long and freq of the DME to be used. I vaguely remember that it required you to simultaneously push 7 & 9 to enter DME mode. Two different ones, ie one in each INS, gave quicker results if available. This system gave an excellent update. You did NOT press DME update because there is no such button/switch on the Delco units, at least those fitted to the B747 classic. Pushing "hold" as you overflew a VOR gave a poor update because you never really knew when you were over the actual VOR transmitter due to the size of the cone at crz altitudes.

[Pepperseed]

Originally Posted by mustafagander

On the Carousel INS, DME updating required you to enter the DME update area of the INS

Greetings

What order of distance are we talking about here ?
Cheers

[FlyingStone]

Originally Posted by Pepperseed

Is the VERIFY POS a message on the CDU ?
May I ask you a quick description of the FMC NAV check Supplementary procedure and where to find it in the Boeing Doc ? I suppose it's all carried out through the CDU ?

It shows on the FMC yes, together with the FMC altert light. The procedure is in the supplementary procedures in the FCOM.