Design

Messerschmitt: Bf109 Ordinates

HughLeave a comment

Messerschmitt: Bf109 Ordinates:

I recently received an inquiry from a museum regarding ordinate datasets for the Bf109. As I previously mentioned in this earlier post the archive I have has a lot of data that was done by others…so I figured it was quite comprehensively covered.

Anyway, I decided to have a closer look and see what was actually documented and how well it aligned with known data. It turns out to be a bit of a muddle. Although the data is quite well presented I have uncovered a number of inconsistencies and anomalies which I will need to resolve.

8-109.000 G-2 Flugzeug Zuss 2

What I thought would be a quick response to an inquiry has initiated a much more intensive study which admittedly I had not planned for. I will go back to basics with this one, using the existing data as a reference and develop a new set of ordinate data for the Bf109 similar to how I approached the Mustang P-51 project.

I have already started with the Horizontal Stabiliser and part of the way through the Vertical. The tables will essentially be a reconstruction from the Messerschmitt drawings and then compiled to develop the X, Y, Z ordinates for transfer to CAD.

bf109

 

Photogrammetry & 3D Scanning

HughLeave a comment

My Thoughts on Photogrammetry & 3D Scanning:

Photogrammetry has come a long way since I first came across it when working for an Offshore Oil & Gas company in the mid 90’s. This company had a department dedicated to researching Photogrammetry and 3d Scanning for surveying existing Process Plant installations.

2019-07-18_18-06-37

Fascinating stuff and it actually worked using predefined targets placed at strategic intervals on the existing plant.

The Aviation restoration companies have for a long time been using 3d scanning and photogrammetry. In fact, if I recall correctly the Smithsonian carried out a survey of the Ta152 which I believe was Photogrammtrey many moons ago. If you look closely at the following image you can see the red dot markers to help ascertain dimensional and positional data from the scans. If anyone out there has access to a copy of the survey results I sure would be interested in seeing them.

Ta 152 H NASM 02

So not willing to fall behind on the latest technologies I endeavored to find out what has been happening with Photogrammetry and played about with some of the available software products. The software I looked at was 3DF Zephyr, Meshroom and Reality Capture. I had a lot of fun but as you may have noted in my previous post the laptop I was using was really not up to the task, nevertheless, I persevered with my research and attained a few good models which surprisingly worked very well.

I even went as far as designing a lightweight support mechanism for photographing fossils in the field. Often these are Sites of Special Scientific Interest (SSSI) in the UK, which is a conservation designation denoting a protected area. Essentially legislation for legal nature/geological conservation of land considered of special interest for their flora or fauna, geology or geomorphology.

Unfortunately, for many of these sites, the removal of fossils is frowned upon so the Fossil Pod was designed to facilitate taking photographs of the found objects…an adapted use for Tensegrity design.

It is early days in my research which I intend to continue to explore options for increasing dimensional accuracy using photogrammetry and obviate the limitations for shiny and reflective objects. I appreciate that the industry may be more interested in 3d laser scanning technologies which is essential for accuracy but I feel that photogrammetry has a lot to offer for presentation and grasping details when a 3d scanner is not available.

As for the software, I would recommend both the 3DF Zephyr and the Reality capture. I should note that all 3 software products do exceptional model generation from photographs however the Meshroom; though in early production; is much less forgiving and requires a very good dataset to work from whereas the others are more tolerant.

If you would like some inspiration for what photogrammetry can do check out this Youtube video from the Swedish Exhibition Agency.

Swedish Exhibition Agency

SEA models on Sketchfab

NAA P-51D Mustang: Landing Gear Dims

HughLeave a comment

NAA P-51D Mustang: Landing Gear Dims

I am currently working on the Landing Gear geometry dimensions to check the data for accuracy. During the course of this research, I thought it may be of interest to share some Excel formula for converting angles shown in Deg Min Sec to decimal degrees and vice versa.

p-51 Landing Gear

The plan angle for the OLEO Strut relative to 25% wing chord is 4° 32′ 35.14″ as shown in the above sketch which translates to 4.543094 degrees.

The accuracy of the angles and dimensions in the NAA documentation is rather good with small deviations occurring of only 0.003mm when developing this in CAD. I should note this deviation is negligible and for all intents and purposes can be ignored. However, I like to get this stuff right so I have set about developing the landing gear dimensions to be as close as possible to be exact.

As I have already developed the cad geometry I measured the same angle above from the CAD system which is now giving me 4.54309673 degrees.

In Excel:

excel table

Starting from the left; in column A; I have input the angle from the cad system, then systematically converted to Deg, Min, Sec in separate cells and then converted back to a decimal angle in column E.

The equations for each cell are as shown below:

p51 Landing gear eq

Just enter the equation in the cells denoted; so for the first equation, this would be in cell B2. The latter equation works fine without parenthesis, which I included just to keep the equation tidy.

The Landing gear geometry will be recorded in a separate spreadsheet and added to the P-51 Ordinate Package. Mustang P-51 Ordinates

P-51D LG

NAA P-51D Mustang: Front Fuselage

HughLeave a comment

NAA P-51D Mustang: Front Fuselage

Following a recent inquiry, I have revisited the NAA P-51 Mustang project to develop the front fuselage and cockpit. This will be an extremely detailed and accurate model including fixings and rivet holes.

P-51D Early.png

The partial structure above is for the early P-51D version; note the instrument panel and support frame. There are also some minor differences with the other components, most notably the curved beam which has different end brackets for the early and late versions.

The 2 instrument panels depicting the P-51D versions, with the later version on the right. This will be a medium to long term project for which I will post regular updates.

Grumman F4F Wildcat: Aileron:

HughLeave a comment

Grumman F4F Wildcat: Aileron:

Having made good progress on the ordinate set for the Grumman F4F/FM2 I decided to put the spreadsheets to one side and do some modeling to verify the dataset. Normally this would not be required to such an extent but I needed to do this to check the relationship between the components and aircraft datums.

I was spoiled with the P-39 project where virtually every component has reference dimensions to the ship center line or thrust lines so positioning was a breeze. However, the F4F drawings sadly lack this reference information on many of the key drawings so developing the 3d cad model is the only sure way to ascertain this data.

F4F Aileron Render

The above model is the left-hand Aileron modeled in Inventor and rendered in Keyshot. Keyshot is a very good renderer, even for a novice like myself; in which you can generate acceptable renderings very quickly. The real-time rendering is very good and will continue without glitches even on a modestly specced system (unlike some of the alternative products). The user interface is logically set out with a good library of materials and textures. I would highly recommend this product.

Getting back on subject; the Aileron ordinates took a long time to complete for various reasons; requiring constant checking and verifying. Once this was done, the modeling was reasonably straightforward except for the small trimming tab. The drawing dimensions are slightly out, so I extracted the neighboring rib profiles to create the template for a finished model.

I also decided to create a few scrap drawing views as a matter of record that will be useful when I eventually move onto modeling the wings themselves.

F4F Aileron 4

For reference; the following image shows the Ailerons attached to the wing assembly. Hinge positions checked and verified with hinge brackets (orange) fitted achieving a planar variation of less than 0.04mm.

2018-10-07_22-29-40

There are still a few items required to complete this model but this is not a priority for me right now. My next objective is to develop the ordinates and perhaps some modeling for the vertical and horizontal stabilizers.

Horizontal Stabiliser & Elevator:

Grumman F4F-FM2 Horiz Stabiliser

F4F Stabiliser

f4f rear fuselage

Tail Fin & Rudder:

F4F FIN RUDDER

F4F FIN RUDDER 01

Fuselage Frame 3:

F4F Wildcat Frame 3a

If you are interested in obtaining my research data for this aircraft then please send me an email. At the moment this is an unfinished project but the available drawings (12) are fully dimensioned which will help you with establishing correct datums and station frame associations along with a few spreadsheets. HughTechnotes@gmail.com

 

Technote: Scaling Ordinates

HughLeave a comment

Technote: Scaling Ordinates:

The primary objective of my work is to record an accurate database of ordinate dimensional data for various aircraft fuselage frames, cowls, wings, cockpit, and stabilizers. This database is derived from manufacturers original documents and drawings.

Often the original source documents are poor quality, occasionally almost illegible, but if we have 95% of the ordinates for a frame then it is relatively straightforward using today’s technologies to determining the missing 5%. Where possible I will cross-reference with part drawings or alternative information to verify.

Cowl nose ring3

However, most archive records are incomplete, as was my frustration with the F6F Hellcat. Having completed the wings, fuselage, and cowl I was stumped by the apparent lack of ordinate data for the tail and horizontal stabilizers (even from part drawings).

There are 2 approaches to determining the missing information. The first is to model the information you do know; from part files, supporting documentation and 3rd part resources. This may provide enough information to determine the missing geometry in order to extrapolate a dimensional data set.

The second; and I would never do this myself; is to trace or convert the outlines of the components from the scanned drawings. There are several products available that will convert raster images to vector files but first, we must achieve a properly scaled image to work with. Most raster image from these archives are scans from 35mm microfilm and due to the nature of the process, the resulting image will not be equally scalable in both X and Y directions.

2018-09-14_14-40-13

Assuming you wish to work with CAD and use this image as a background I would recommend the following process to achieve the best result. This particular drawing is created from actual loft templates and includes the locating pins set to a specified distance in each of the corners plus a drawing scale rule.

Some drawings may only have scale rules, either way, the process is the same.

If we insert this image directly into a drawing in Autocad or similar the only option is a user-defined global scale parameter which will scale the image equally in both X and Y directions, which is not what we want. Even once the image is inserted the option is the same.

The best way to circumvent this is to insert the image into a drawing, without any scale parameters applied. Then save this drawing including the image as a DWG file.

Xref this drawing into another drawing and you will be presented with the following dialogue box ( I am using Draftsight but Autocad will be similar).

2018-09-14_15-30-28

As you can see you now have the option to apply different scales to the X and Y directions. This works very well and will provide a very good reference for your work. I should clarify that some CAD products have the option to insert an image as an Xref but the scaling options are not the same as for a DWG file, instead reverts to a global scale option only.

As a workaround for missing information, this is a very accurate way of achieving a good result and will satisfy the majority of applications.

As my projects are records of known dimensional information this process would not be applicable.

Grumman F6F Hellcat: Cowl Ordinates

HughLeave a comment

Grumman F6F Hellcat: Cowl Ordinates:

I may have been a tad over-optimistic in my previous article on the F6F ordinates when I mentioned there was a good chance the Grumman ordinate data would be complete. The one exception is the cowl, although there is an ordinate drawing for the cowl the table itself is just a black blob, completely illegible. However, I checked the part drawings that make up the cowl which is quite well detailed so I made a start on the nose spinner ring.

The spinner ring main model is fine and here it is derived into the air scoop construction model. The construction requires numerous contour lines as each ordinate needs to be manually checked due to the poor quality of the original Grumman drawing scan.

Cowl nose ring3

This ordinate drawing is not deliberately blurred by me, this is how it actually looks like. The main dimension is almost legible just the fractions that are problematic. The process will require evaluating each contour line and curvature check, both horizontal and vertical as shown in the above image.

I am hopeful of achieving a good result with this model which will probably take a few days to complete.

Update 27th August:

F6F Cowl Nose Ring2

I have now determined the correct ordinates for each of the seven profiles. Notice the lower profiles have been artificially extended to the center ring, which will give me better results when lofting. After lofting a surface the plan then is to remove the mouth of the air scoop (blue) and apply the finishing flange to the inner edge.

F6F Cowl Nose Ring80

The ordinates are recorded in a spreadsheet with the x,y,z coordinates extrapolated as I previously did for the wings and fuselage.

Update: 12 Sept 2018:

Cowl Ring Cowl Nose setout dimensions verified.

F6F Ring Cowl Nose

Ordinate Dataset Completed: Wings, Fuselage and Front Air Scoop.

The F6F archive of scanned Grumman documents comprises over 7000 drawings in PDF.

Grumman F6F Hellcat: Wing Ordinates

HughLeave a comment

Grumman F6F Hellcat: Wing Ordinates

I wrote an earlier article on my work with the F6F fuselage ordinates which I have recently updated. Whilst revisiting the archive I also came across the Grumman wing ordinate drawings and decided to develop those as well. I was reluctant to do this as the original drawings were not that great.

Thankfully it was not as bad as I first suspected, though it has taken me over 7 hours to painstakingly enter each ordinate manually to tabulate the ordinates in Excel.

F6F Wings

I still have to interpolate the data to generate the appropriate X, Y, Z coordinates; set out from the 35% chord; which I will endeavor to do over the next few days.

F6F wings 2

To verify the ordinate dimensions the following equations are applied. The chord length is for any wing chord whilst the LER is only applicable from station 75 to station 252.

f6f calcs2

To be honest the F6F Hellcat was not even on my to-do-list but a conversation with a colleague about the F6F performance characteristics prompted me to have a closer look at the archive. Surprisingly it is very possible that this archive may have sufficient information to generate an entire aircraft ordinate set, which is quite rare.

f6f

I will update this post when the wing model is complete, so come back soon.

Update August 23:

 

I have checked the Centre section profile for accuracy and noticed one point out of alignment by 2mm towards the leading edge. Removing this point allowed the natural curvature of the spline to define an acceptable profile as shown. The curvature check shows that this curve now matches the Leading Edge Radius.

The trailing edge extends beyond the 100% Chord by 5/8th inch on the centre section (Station 0) which tapers to zero at Station 252. Drawing a straight line segment from the Trailing Edge Radius results in perfect alignment with the spline.

Centre Section Stations:

f6f ctr section

Outer Panel Stations:

F6F outer wing

It is not unusual to have a few rogue points from the tabulated ordinate data which is why it is important for a detailed analysis like this.

And here, at last, the complete wing assembly:

f6f wing assembly

Vought F4U Corsair: Ordinate Data

HughLeave a comment

Vought F4U Corsair: Ordinate Data;

About 6 months ago I received an ordinate dataset from a good friend Gary Henry for information purposes. It is a very comprehensive set for the fuselage comprising over 2800 points to define in excess of 1930 individual ordinates.

Vought_F4U_Corsair_(USMC)

I have recently updated my data processing procedure utilizing new features in MS Excel particularly the “TEXTJOIN” command which makes it a lot easier to extrapolate the X,Y,Z ordinates from large datasets. This dataset was ideal to work with the new process.

F4U Corsair Ordinates

The Textjoin function allows you to predefine a delimiter and then select either an array of data or individual cells using the Control/Mouse combination. You can see I have locked in the selected column and the top row. The units shown are inches but can easily be converted to millimeters.

F4U Corsair RevB

Due to the nature of the dataset, there is a very distinct central plane on the zero vertical plane, which of course I would filter out if I decided to progress this further as a CAD model. I don’t have enough of the manufacturers’ original drawings to develop this aircraft at this time but it sure is interesting working with other datasets.

F4u-1 sideview

The dataset is actually very good only 3 points not quite in alignment. I profiled the top and bottom contours and the contours either side of the fuselage centerline; all 4 curvatures were very smooth.

Update Oct 2018:

Recently received some new data that has allowed me to progress this project with the development of the Cabin ordinates as shown below.

F4U Cabin

F4U Corsair2

HiRise data and WRL Conversion.

HughLeave a comment

HiRise data and WRL Conversion:

It has been a while since I last posted an article due to being busy with other projects. During some research activites, I came across a number of subjects that may be of interest, two of which I would like to share.

The first one is the HiRISE Digital Terrain data models on the University of Arizona website. The website contains datasets that are digital extractions of surface terrain scans of the planet Mars. The DTM datasets are publicly available for research and modeling of geological processes.

PSP_007100_1520

Naturally curious I decided to investigate the possibilities of modeling and rendering of these datasets from which I produced a few preliminary 3d terrain models using Blender and rendered in Keyshot…Gorgonum Chaos:

mars10The technique I used is described in this video on Youtube, clearly explaining the process. To me, this is incredible stuff and thanks to the University of Arizona for all their dedicated work in developing these datasets. So have a look and check it out for yourselves.

The next subject is WRL. WRL is a file extension for a Virtual Reality Modeling Language (VRML) file format often used by browser plug-ins to display virtual reality environments. VRML files are known as “worlds,” which is what WRL stands for.

One of my many interests is Tensegrity, a structural form of tension and compression members first developed by a chap called Kenneth Snelson. The internet is full of examples of this structure concept inspiring many variations from fairly simple to very complex designs. I have developed a few of my own.

delta

Many of the practitioners in this field make datasets available for personal use and one particular format they use is the VRML (WRL) so you can view the design in 3d.

For the simple structures similar to this image the design and construction are not that difficult, however when it comes to developing your own version of the more complex examples it can be a real headache. Although some datasets include actual point cloud data the process of matching pairs of points to reconstruct the design can be a nightmare.

The obvious solution would be to convert the WRL model into something usable that could be used as a guide for developing a 3d cad model. I tend to favor Meshlab for doing this as it is one of the few programs that will accurately convert the imported data.

meshlab

The WRL model is converted into a series of mesh objects that we can export as an OBJ or STL file and then import into Inventor.

Once in Inventor, it is simply a case of selecting each of the compression struts and “Fit Mesh Face”. Select the “Auto Fit” option for each member and it will create a surface from each mesh representing the struts.

The tension wires are then created as a 3d sketch using the background mesh model as a guide. At this stage, the model is a workable composite but may require micro adjustment for the tension wires to ensure the finished item is properly constrained. I would reverse engineer this model and reconstruct as an assembly and apply the microdimensional adjustment to the groups of tension wires to ensure the absolute accuracy of the final design.

It is beyond the scope of this article to go into the detail of every step, but if you require information on any of the topics please feel free to drop me a line.

Tensegrity Conv

I hope you find this article interesting and have fun.