Featured Project Azm: Pakistan's Ambitious Quest to Develop 5th Generation Military Technologies.

[JamD]

@Bilal Khan (Quwa) looks like Turkey’s aviation city is off the ground. How long do you reckon Pakistan will complete its own aviation city at Kamra?

Turkey's Aviation City vs Pakistan's Aviation City:

Just my guesses. And also here's some context:

In simple words they jammed an axe into their foot because if lack of long term vision...
The request for funds is understandable as the prices for land acquisition have spiraled from 10s of thousand to millions...
At first they were using middle men in the premise that the land acquisition is for a factory for building helis. this is during the time the T-129 talks were in full swing and acquisition seemed to be in good numbers... So people were ready to give away land at marginally higher prices and the middle man ofcourse made his commission....
Someone in the higher ups decided that they should eliminate the middle man altogether. This went really badly for them. Now they are holding large chunks of land to this day in between the acquired area. This makes it almost impossible for making a functioning part of the project...
Subsequently prices have spiraled to a new high... ( 50k per kanal to as high as 12-45 lac )..... The project cost excluding inflation has sky rocketed.. People are holding there land for better prices and land acquisition is now at a sluggish pace... As of 3 month back they had acquired almost 6k to 8k kanals while if my memory serves me right the whole project is of 25k kanals.. so you can see the disparity...
You have stated that there is a lack of R&D related jobs and the reason you have stated yourself.... Only buildings raised in the newly acquired area is the university and adjacent hostel....
The head of land acquisition isn't that in touch with the locals either.
All in All I don't see the project completing in the next 10 years if the govt doesn't use section 4.
( Corruption, bribes and other arm bending tactics have been left out )

No, even Army now has a system of acquisition and secrecy with these kind of things.... They have a serving wing commander running the sight office, sigh if anyone of you is actually connected high up please try to make them understand that this is a very social job and a person who has to run something like this needs only 2,4 years to make necessary relations with Tehsildars, Patwaris and the locals... Most of these lands when they were starting the project were held by families and acquisition was easy before... Now there are investors..
Only way forward seems to me is the shutting down of one of the older Mirage or F-7 factories.... This also helps in understanding why J-10s are ordered ( retirement of one of the two ).

 

[JamD]

If I am not mistaked @JamD or some other person mentioned that PAC gave that control systems lead engineer guy 6 months to make a UAV out of K8. But he did not agree to it due to limited time/budget. If it was me, I would have taken the project regardless of the probability of success. At worst, we will have a crashed aircraft, but at best, we will have a workbench for further prototyping and development.

Such seemingly-impossible projects act as rapid capacity building in control systems/avionics. Building a first iteration of the FCS (with limited functionality) would have been accomplished. If the engine does not have it already, the first version of FADEC could have been developed. Small measurable consistent incremental progress. In my current job, we plan to do exactly that but to compete against european aerospace industry giants.

I would have hired 5 control engineers with 3-4 embedded SW and HW developers. Control SW could have been written using any free RTOS with a simulation framework (HiL/Pure SW) until HW was fully ready. A very popular FCS framework for commercial/DIY drones already exists online. Major work would have been control algo, and HW components (PCB, actuators, etc).

Not a K8, a Mushaak. And I don't know if that was the reason he left or not. Probably many other reasons.

The FCS knowledge at AvRID/PAC is extremely limited - you have to keep in mind they are starting from 0 with no help/communication with SPD organizations who do have a lot of experience with FCS. Compartmentalization is a big issue. However, I am positive about the environment of AvRID/PAC as it is much less suffocating and more conducive to research than SPD orgs. If AvRID survives long term it will impress all of us I believe.

 

[Muhammad Saftain Anjum]

Not a K8, a Mushaak. And I don't know if that was the reason he left or not. Probably many other reasons.

The FCS knowledge at AvRID/PAC is extremely limited - you have to keep in mind they are starting from 0 with no help/communication with SPD organizations who do have a lot of experience with FCS. Compartmentalization is a big issue. However, I am positive about the environment of AvRID/PAC as it is much less suffocating and more conducive to research than SPD orgs. If AvRID survives long term it will impress all of us I believe.

What about FCS of Azam MALE UCAV?
Is it indeginous or imported?
S2 has probably indeginous FCS.

 

[JamD]

What about FCS of Azam MALE UCAV?
Is it indeginous or imported?
S2 has probably indeginous FCS.

Don't know. The high speed target drone was using pixhawk (https://pixhawk.org/) last I checked.

S2 probably since its SPD.

 

[Al_Muhannad]

Not a K8, a Mushaak. And I don't know if that was the reason he left or not. Probably many other reasons.

The FCS knowledge at AvRID/PAC is extremely limited - you have to keep in mind they are starting from 0 with no help/communication with SPD organizations who do have a lot of experience with FCS. Compartmentalization is a big issue. However, I am positive about the environment of AvRID/PAC as it is much less suffocating and more conducive to research than SPD orgs. If AvRID survives long term it will impress all of us I believe.

Compartmentalization is a tricky thing- one need it for security but at the same time it is inefficient and kills flow of ideas.

 

[The Eagle]

I don't know if that was the reason he left or not

If you can't be corrected without being offended, you will never grow in life. Should have tried.

 

[Al_Muhannad]

The high speed target drone was using pixhawk (https://pixhawk.org/) last I checked.

hypothecially speaking, if you were to do the algo side of FCS, how long and what type of resources would you need?

 

[JamD]

hypothecially speaking, if you were to do the algo side of FCS, how long and what type of resources would you need?

1. Well instrumented wind-tunnels of various sizes and various speeds.
2. A model-building shop that can make models for wind-tunnel testing and scale-flight testing. This place would need 3D printers, and skilled R/C aircraft makers, machinists. Might get a metal sintering 3d printer if I have the resources to make models for windtunnels:
   
   
   
   This shop will probably need 20ish technicians.
3. Hire 10ish people who specialize in CFD and 3-4ish people that specialize in reduced order model. The idea is that computer models are built, tested using CFD and reduced order models (ROM), and some of these tests are replicated in wind tunnels and R/C scale flights to get actual data to correct the computer models. Then these computer models can be run millions of times for design of control systems.
4. A high-performance computing facility to run all of this CFD code.
5. Hire a 10-20 people team to work on embedded systems - I don't know if I'll need FPGA but would definitely consider it.
6. Hire 5-6 experts on sensors like gyros, accelerometers, air-data systems, data-driven estimation.
7. Hire a 10-20 people team of coders to write and test the actual flight software. Testing on models and R/C scale models.
8. Hire 3-4 people that specialize in hyrdraulics and pneumatics and motors. I want a few people to only look at actuators.
9. EDIT: Forgot the most important part lolol:
   I will hire 5-6 experts in various kinds of control: classical, optimal, robust, adaptive, model-predictive. This is so that I have all kinds of options for flight control laws.
10. Hire 2-3 system engineers that can help us put all of this hardware and software together.

Given a project I can probably get a basic FCS system in 2 years maybe? I know you said "algo side" but I don't think you can neatly separate the algorithm side since the algorithm wholly depends on the rest of the pieces in the ecosystem.

Also major major disclaimer: obviously I have never been the head of flight dynamics and control for an aircraft so this is my VERY VERY rough guess based on subject matter knowledge. Take it all with a massive saltmine worth of salt.

If you can't be corrected without being offended, you will never grow in life. Should have tried.

I would've tried too, but I don't know the other circumstances so I don't want to pass judgment. The incident felt like he was shot down because he asked for more time and his bosses were saying that the task was only worth 6 months of effort. I can see how that is an extremely toxic environment to be in: your boss - who isn't the expert on controls, you are - tells you that you are incompetent at your job and if he were doing it he could do it better. Hard to resist the urge to respond with, "good, have a go then!"

 

[Al_Muhannad]

1. Well instrumented wind-tunnels of various sizes and various speeds.
2. A model-building shop that can make models for wind-tunnel testing and scale-flight testing. This place would need 3D printers, and skilled R/C aircraft makers, machinists. Might get a metal sintering 3d printer if I have the resources to make models for windtunnels:
   View attachment 810543
   This shop will probably need 20ish technicians.
3. Hire 10ish people who specialize in CFD and 3-4ish people that specialize in reduced order model. The idea is that computer models are built, tested using CFD and reduced order models (ROM), and some of these tests are replicated in wind tunnels and R/C scale flights to get actual data to correct the computer models. Then these computer models can be run millions of times for design of control systems.
4. A high-performance computing facility to run all of this CFD code.
5. Hire a 10-20 people team to work on embedded systems - I don't know if I'll need FPGA but would definitely consider it.
6. Hire 5-6 experts on sensors like gyros, accelerometers, air-data systems, data-driven estimation.
7. Hire a 10-20 people team of coders to write and test the actual flight software. Testing on models and R/C scale models.
8. Hire 3-4 people that specialize in hyrdraulics and pneumatics and motors. I want a few people to only look at actuators.
9. EDIT: Forgot the most important part lolol:
   I will hire 5-6 experts in various kinds of control: classical, optimal, robust, adaptive, model-predictive. This is so that I have all kinds of options for flight control laws.
10. Hire 2-3 system engineers that can help us put all of this hardware and software together.

Given a project I can probably get a basic FCS system in 2 years maybe? I know you said "algo side" but I don't think you can neatly separate the algorithm side since the algorithm wholly depends on the rest of the pieces in the ecosystem.

Also major major disclaimer: obviously I have never been the head of flight dynamics and control for an aircraft so this is my VERY VERY rough guess based on subject matter knowledge. Take it all with a massive saltmine worth of salt.


I would've tried too, but I don't know the other circumstances so I don't want to pass judgment. The incident felt like he was shot down because he asked for more time and his bosses were saying that the task was only worth 6 months of effort. I can see how that is an extremely toxic environment to be in: your boss - who isn't the expert on controls, you are - tells you that you are incompetent at your job and if he were doing it he could do it better. Hard to resist the urge to respond with, "good, have a go then!"

I would say for anything software (embedded, FSW, testing) 15 people are required at MAX if they have a good work ethics/culture. 15 include a 50% margin (I would try even with 10/5(initially) people who are super motivated with no sarkari job attitude or only-here-for-money attitude). What kind of package does PAC offers for contract jobs?

My MS was 30% Innovation and Management. One key takeaway was entrepreneurs do not require dedicated office spaces/equipment/non-essential personnel. My current company has <15 SW people, <15 HW (who are specialized in PCB, harness, power etc), and <15 system engineers. You would not believe what our aims are (we operate in market similar to PAC). Everything that is non-essential will be done later, or leased. We did not have even good internet, or toilet cleaning service till last month.

What would be the first steps? like technical milestones with priority order. I only took one control course, and had no further exposure to it since then. When I said "algo side", I meant that whatever the final algorithm is implemented using RTOS on an embedded device which may use FPGA. To RTOS, it just a periodic computation done at the sampling rate of the digital control algorithm (please correct me if I am wrong).

 

[Al_Muhannad]

What sort of financial, institutional support once can expect from PAF/PAC/Govt/Private sources?

 

[JamD]

I would say for anything software (embedded, FSW, testing) 15 people are required at MAX if they have a good work ethics/culture. 15 include a 50% margin (I would try even with 10/5(initially) people who are super motivated with no sarkari job attitude or only-here-for-money attitude). What kind of package does PAC offers for contract jobs?

My MS was 30% Innovation and Management. One key takeaway was entrepreneurs do not require dedicated office spaces/equipment/non-essential personnel. My current company has <15 SW people, <15 HW (who are specialized in PCB, harness, power etc), and <15 system engineers. You would not believe what our aims are (we operate in market similar to PAC). Everything that is non-essential will be done later, or leased. We did not have even good internet, or toilet cleaning service till last month.

What would be the first steps? like technical milestones with priority order. I only took one control course, and had no further exposure to it since then. When I said "algo side", I meant that whatever the final algorithm is implemented using RTOS on an embedded device which may use FPGA. To RTOS, it just a periodic computation done at the sampling rate of the digital control algorithm (please correct me if I am wrong).

This will become very technical very fast. But you are right that you obviously need to enforce hard real-time. Your algorithms need to be preferably fixed-complexity and designed with digital control in mind (not just continuous-time laws approximately discretized). For most applications you will not use very fancy algorithms (just use digital PID) so FPGA's might be very useful. There will be a lot of software testing that I am obviously not an expert in (I have some experience working in an electric vehicle startup where they were writing code for the inverter and battery management system, which is the extent of my knowledge). My research focus is on real-time implementable control laws so I have a tendency to get too technical too fast. It's best not to go there.

The first priority would be getting good, verifiable models. Parallel first priority, get basic hardware flying that includes the ability to test basic control laws. Everything else can wait.

What sort of financial, institutional support once can expect from PAF/PAC/Govt/Private sources?

For what?

 

[Mig hunter]

Let me emphasize, no 5th Gen aircraft is going to be manufactured by Pakistan, at least not in near future keeping in view the capacity, funds and technological and manufacturinh base.. Pakistan will either buy off the shelve or Co developer just like thunder..

 

[Al_Muhannad]

This will become very technical very fast. But you are right that you obviously need to enforce hard real-time. Your algorithms need to be preferably fixed-complexity and designed with digital control in mind (not just continuous-time laws approximately discretized). For most applications you will not use very fancy algorithms (just use digital PID) so FPGA's might be very useful. There will be a lot of software testing that I am obviously not an expert in (I have some experience working in an electric vehicle startup where they were writing code for the inverter and battery management system, which is the extent of my knowledge). My research focus is on real-time implementable control laws so I have a tendency to get too technical too fast. It's best not to go there.

The first priority would be getting good, verifiable models. Parallel first priority, get basic hardware flying that includes the ability to test basic control laws. Everything else can wait.


For what?

Yup I was thinking exactly that. First major milestone should be this modeling for the control algorithm. Software can be changed comparatively fast as compared to the model, and gets mature as the project moves.

I was asking about investment if such a mechanism exists to support engineers on new ideas in aviation/EW.

Let me emphasize, no 5th Gen aircraft is going to be manufactured by Pakistan, at least not in near future keeping in view the capacity, funds and technological and manufacturinh base.. Pakistan will either buy off the shelve or Co developer just like thunder..

5th gen aircraft is too big a target. Let's get a indigenous FCS done first.

 

[Trango Towers]

Yup I was thinking exactly that. First major milestone should be this modeling for the control algorithm. Software can be changed comparatively fast as compared to the model, and gets mature as the project moves.

I was asking about investment if such a mechanism exists to support engineers on new ideas in aviation/EW.


5th gen aircraft is too big a target. Let's get a indigenous FCS done first.

Think small achieve small

 

[Al_Muhannad]

Think small achieve small

I was talking about milestones, not roadmap.