Note: this is not a complain thread. This thread is meant to raise a point, and to give constructive feedback. Above all it is meant as a question to PMDG.

Hello captains,

I have been on 737NG's many times. Once, I even had the chance of having the unforgettable experience of sitting on the jumpseat of a 737-700 during take-off.  In all of my flights, I always noticed something about the initial engine spool-up to 40% N1 for take-off. It always seemed to me, by listening to the engines, that they got some sort of a "boost" at around 30% N1, and then stabilized instantly.

As soon as the PMDG 737NGX got released, I immediately noticed this fundamental difference. It seemed to be taking ages for the engines to stabilize, especially at lower N1 settings. So I made a thread, raising this point. Everybody immediately started to shout to me that I was wrong, because I didn't have access to data, so I must be lying. (I must admit, my explaining skills back then weren't as good as they are now) Two flights on the 737NG lately, once again confirmed my opinion.

So this time I thought, OK, I will raise this point again now, this time giving the best evidence and data I can get my hands on. In this way, I will try to show you a proper comparison between the engines spooling up to 40% in the PMDG 737NGX versus the engines spooling up to 40% in the real-world.

First, let me show you a video on YouTube, of a KLM 737-900 taking off from Amsterdam Schiphol Airport, runway 24. In this video, skip to 1:05, where the captain sets initial thrust to around 45% N1. Please observe the N1 readout on the EICAS

Comparing this, to spooling-up the engines to the same 45% N1 in the PMDG 737NGX, which differences can we notice? If you look closely at the video, you can see that the acceleration of the engines in the beginning, is very very low. Then however, acceleration starts to increase gradually until the engines get a HUGE boost at 35% N1, making the engines instantly spooling up to 45% N1, then they hop back a few percent, and are then stabilized. 

In the PMDG 737NGX, the acceleration of the engines in the beginning of the spool-up sequence, appears to be quite a bit higher. Then however, acceleration starts to decrease gradually, and from then on it takes aaaages before the engines are properly stabilized. In other words, it seems a bit like the complete opposite.

Note: in this case I am defining "stabilized" as "N1 not changing by more than 0.1% a second". 

To give a better image of my findings, I conducted a simple - not entirely accurate - experiment and put the results into a graph. As said, it's not a 100% accurate, but numbers should not be off by more than 1% of N1. 
The way I conducted this experiment, was by using a stopwatch on my smartphone, and then, from the moment the captain set thrust to 45% N1, every second I paused the stopwatch while simultaneously pausing the video. Then I wrote down the N1 readout plus the time on the stopwatch belonging to it. I did this until the engines were stabilized. I conducted this experiment 5 times, and then took the average of each of my results. Each time, results did not differ by more than 1% of N1, so it should give us a good global image of the spool-up sequence. Next, I did the same thing inside FSX using the PMDG 737NGX.

Here are the results I found:

 

 

These graphs I made, pretty much match up with my findings and experiences. If we analyze this graph, we can see that initially, the acceleration of the engines in the NGX, is ahead of the acceleration of the real-world engines. However, then the acceleration of the real-world engines, starts to increase gradually, and then, at around 35%, gets a boost to 45% N1, then they hop back a little to 44%, as you can see by the red line dropping down again. Next, the engines are stabilized at t = 8.0 seconds. What is also very noticeable, is that the engines in the NGX at this moment, are still finishing to stabilize. Only at t = 11.0 seconds, we can see a complete horizontal line. 

 

In the thread I made around two years ago, a real-world pilot claimed the engine spool-up sequence in both the real-world aircraft and the PMDG NGX are pretty much the same, because the spool-up time to 40% N1 is in both cases the same. As he is a real-world pilot, I believe him. And it turns out, he is right about this statement, because the in the graph the green and red line are crossing eachother at around 40% N1 meaning that the spool-up time up to that point is in both cases the same. But he forgot to mention however, is how the engines spool-up within and after this moment. As it turns out, quite a bit differently. 

What's actually bothering me about this issue, is that in the NGX I have to wait quite a bit longer for the engines being completely stabilized, while in the real-world TO/GA can already be engaged after 8 seconds. 

Am I being very perceptive here? Well, yes, I am a perceptive person, and I always look for details. However, considering that PMDG takes detail to a 100 times higher level, I might not be so perceptive after all. 

 

So, my question to PMDG is. Why is this aircraft so perfect, because really, it is, isn't it? :Big Grin: I love it to death, but I never understood why this very obvious engine behaviour, was never modeled in the NGX? Was it simply an FSX limitation? Was it a lack of proper data not allowing you guys to model this? Or was it something else?

If it's simply an FSX limitation, which I suspect it is, I can easily live with it. Whatever it is, it won't stop me from using your product, I still love it as much as I did since the beginning. 

 

 

How are you controlling for your application of thrust between 0 and 45%?

 

If the time you're trying to study is between 0 and 45%, which is where the pilot is the one working the throttles, you've opened yourself up to variation.  Unless you've mimicked exactly what he does in the video (tough because he begins movement prior to the camera panning down), then what you're noticing as a difference could easily be written off as human variation.

 

Looking back at all of my real world flying videos, I apply takeoff power differently each time (depending on the field, how much attention I'm giving to adding throttle, and so on).

How are you controlling for your application of thrust between 0 and 45%?

 

If the time you're trying to study is between 0 and 45%, which is where the pilot is the one working the throttles, you've opened yourself up to variation.  Unless you've mimicked exactly what he does in the video (tough because he begins movement prior to the camera panning down), then what you're noticing as a difference could easily be written off as human variation.

 

Looking back at all of my real world flying videos, I apply takeoff power differently each time (depending on the field, how much attention I'm giving to adding throttle, and so on).

Kyle, please note here that I am not studying the engine spool-up defined by the percentage of throttle input, but I'm studying the engine spool-up by the percentage of N1. In this case, between 20% (idle) and 45% (initial spool-up before TO/GA). Unfortunately when I drew my graph, I didn't know how to show only ranges from 20 to 50 instead of 0 to 50, hence the reason I drew a dotted line, so the range from 0-20% is supposed to be ignored here.

 

Also, I don't think the "way" of setting thrust has that much of an effect in this case because it takes 8 seconds for the engines to spool-up, and you can see the captain has the throttles set stable from the moment the engines were still in the low 20's of N1. So this boost would be expected if you would increase the throttles quickly, at let's say 30% N1, then pulled them back again at 45%. This method is also my way of cheating on the engines in the NGX to make them stabilize quicker.

 

EDIT:

 

Kyle, for your interest:

 

 

Right during the moment the camera was panning down I paused the video, and no thrust was set yet by the captain.

 

After that, the captain set throttles to 45% N1 and never made a change until stabilization. In other words, I mimicked exactly what he did, as this was the whole point of my comparison.

 

Even if a different way of setting thrust would influence the way the engines spool up, the the difference should not be THIS big, right? I tried different ways of setting thrust, and the only difference I could notice was a quicker spool up on average, but certainly not magical sudden boosts of engine acceleration or different amounts of acceleration at different moments.

 

 

Kyle, please note here that I am not studying the engine spool-up defined by the percentage of throttle input, but I'm studying the engine spool-up by the percentage of N1.

 

...but what do you think drives engine % N1?  If I move the throttle levers slowly from idle to 40% by throttle position alone, the N1 will show a similar slow trend.  If I move them slowly, and then quickly, N1 will follow that.  Say I move them slowly from 25% to 30% and then move them quickly between 30% and 40%.  The engines aren't going to average that movement out; rather, since the delivery of the fuel is driven by the movement of the throttle (when I moved the throttle levers), fuel was slowly delivered up to 30% and then more quickly delivered thereafter.  As such, the engines will slowly accelerate between 25 and 30%, and then quickly thereafter.

 

Unless I'm mistaken (I don't fly the thing, and I didn't design it), your direct input on the throttles and rate at which your inputs are made will affect N1 in the same way.  The only reason it seems somehow separated is that there's lag in between throttle movement and N1 movement, simply in the nature of the jet engine.

 

Again, unless there's hard evidence to show otherwise, I still stand on the seat to throttle interface being the cause.

 

 

(1) Above graph is a Classic Step response   to step input ( Blue input)

 

(2) What you graph shows is a result of a more linear input , as throttle is move over a finite time.

 

(3) With the non-step input, one would get ( to  a first approximation) , graphs very similar to what you have.

 

(4) The Difference now on your graph, is the loop gain  K.

 

It would appear that PMDG have too low a gain ("under damped" - "no overshoot") , and the Real world, is as expected, optimized to "Critical damping",(ie  slight Overshoot)  gain of Kp, Ki, & Kd, which results in reaching the desired stable, end point, in the minimum of time.

 

This is only  first approximation,, There is obvious a lot more going on in this "Control System",  and the way the Real World Engine Management Computer operates, but the major difference you are seeing between the PMDG, and "your" real world measurements, are most probably  attributable to the Loop Gains K.  ( Kp, Ki & Kd)

 

====

 

BTW: A very impressive piece of "In the Field" measurement.

Also, not easy to get to be a spectator in a cockpit during takeoff, since 911.  Gone are the "good old days"  :(

 

 

(1) Above graph is a Classic Step response   to step input ( Blue input)

 

(2) What you graph shows is a result of a more linear input , as throttle is move over a finite time.

 

(3) With the non-step input, one would get ( to  a first approximation) , graphs very similar to what you have.

 

(4) The Difference now on your graph, is the loop gain  K.

 

It would appear that PMDG have too low a gain ("under damped" - "no overshoot") , and the Real world, is as expected, optimized to "Critical damping",(ie  slight Overshoot)  gain of Kp, Ki, & Kd, which results in reaching the desired stable, end point, in the minimum of time.

 

This is only  first approximation,, There is obvious a lot more going on in this "Control System",  and the way the Real World Engine Management Computer operates, but the major difference you are seeing between the PMDG, and "your" real world measurements, are most probably  attributable to the Loop Gains K.  ( Kp, Ki & Kd)

 

====

 

BTW: A very impressive piece of "In the Field" measurement.

Also, not easy to get to be a spectator in a cockpit during takeoff, since 911.  Gone are the "good old days"  :(

Interesting, any idea why the loop gain effect wasn't modeled in the NGX?

I have no insight as to what PMDG did or didn't do,  but I suspect they matched the 6.5 seconds to get to 40% N1 time, as you so clearly showed in your graph, and as you describe, was confirmed as being "realistic" by a real world 737 captain at some time in a post.

 

Maybe, due to the limitation is FSX modeling of the engine,  if you adjust those airfile parameters that are availabe to you, so that "time to 40% N1" is  correct, you have to live with the resulting overall response, as coded within FSX's modeling.

 

Noby has ever claimed that FSX's turbine modeling was that great !!

 

However, all wild guesswork on my behalf, based on what evidence you have presented, and a wet finger stuck up in the air.  :dance:

IBTL

While not necessarily writing off your point, I would add that atmospheric conditions play a part in engine response. Unless you can be sure to match these, it's not a fair test.

While not necessarily writing off your point, I would add that atmospheric conditions play a part in engine response. Unless you can be sure to match these, it's not a fair test.

I just did the test again in the NGX with this time using the same atmospheric pressure of 1004 HPA and the same outside airtempature of 10C. No difference in spool-up time or any other differences were noticeable. 

 

Now, I do think you are right that atmospheric conditions affect the engine's spool-up rate, but I think it will only have a linear effect. I.e. in my graph the red line should have a similar shape to the green line, but then either steeper or less steep. In this case the first part of the spool-up until 30% N1, can be caused by a difference in atmospheric pressure. However, I don't believe that this spike that is visible on the red line, is magically caused by a difference in atmospheric conditions. This spike was, by FSMP, described as an overshoot in engine thrust to make the engines stabilize quicker at the position the thrust is set at, the so-called "loop gain". This loop gain is from my personal experience always present, as I have always heard it on any of my flights on the 737NG.

 

 

Arjen, you are forgetting the purpose of this forum. THERE ARE NO BUGS IN THIS PRODUCT. This product is perfect and there can be no improvements. 

Arjen, you are forgetting the purpose of this forum. THERE ARE NO BUGS IN THIS PRODUCT. This product is perfect and there can be no improvements. 

 

What what what about....................weather radar

 

 

If you don't have a sense of humor please well...don't reply

I wonder if the phenomenon you're witnessing in real application is due to the resolution of the monitoring equipment or the refresh rate of the display.  That could explain the rapid rise from 30 to 40%.  Just a thought.

 

Mike

I wonder if the phenomenon you're witnessing in real application is due to the resolution of the monitoring equipment or the refresh rate of the display.  That could explain the rapid rise from 30 to 40%.  Just a thought.

 

Mike

No,

 

The reason why I made a comparison in the first place, was because I had already experienced and heard this "quick rise" in engine RPM on all of my flights on the 737NG during take-off.

 

Several YouTube videos, and the graph I made exactly confirmed these findings (for me).

 

I don't believe the connection between the two is just coincidence...

 

 

I just did the test again in the NGX with this time using the same atmospheric pressure of 1004 HPA and the same outside airtempature of 10C. No difference in spool-up time or any other differences were noticeable.

 

Did you match relative humidity and density?

 

Aaron

-oOo-