NAA P-51D Mustang: Tail Wheel Progress Update
I’ve been busy building the 3d models and working through the vagaries of the Mustang P-51 Tail wheel assembly. This is a selection of the new parts built to date:
This last image is a photo showing the Housing for the Tail Wheel Spindle with the model in a similar orientation and finish for comparison.
For the latest version of these models please refer to this article.
NAA P-51D Mustang: Tail Wheel Housing EndCap
This item #91-34005 is the cap to be fitted to the end of the tail wheel housing described in the previous post.
The drawing that this was based on details the finishing and machining from a forging. Unfortunately and is now seeming to be quite typical for the tail wheel area again I am missing key information…its not that the drawing is lacking the necessary detail for its intended purpose as it is based on an existing unit its just that I don’t seem to have the existing forging drawing as a reference.
The missing information; relating to diameters and material thicknesses; would have been on the forging drawing and the draughtsman did not really have any cause to replicate this information on his finishing drawing…though it would have been enormously useful!
However this is not always the case as depending on the draughtsman some do include this type of information as reference dimensions.
I have therefore interpolated the missing details as best I can based on the information available and where possible cross referenced against other drawings.
The forging drawing is actually listed in the NAA register but the archive record lists the wrong reference drawing so its likely that there is a copy of this forging somewhere in the archive!
For my initial purposes though I think the data is sufficient for the end goal of creating a working tail wheel assembly as part of the mechanism study.
NAA P-51D Mustang: Tail Wheel Housing
In my endeavors to develop the assembly and parts register I have made some progress determining the associations in conjunction with the listing in the NAA documents. I have started with the Tail Wheel assembly,which actually has 59 drawings for the sub assemblies and parts…it was surprising the number of drawings just for this one area!
This is a partial screen shot of the assembly register; again divided by sub-assembly level down to the individual parts.
Having collated this information I started work on building the 3D Cad models, starting with the Tail Wheel Oleo Housing.
This housing is perhaps the most difficult model in the series for the Tail Wheel and it has given me problems.
The main lower bracket struts are shown in the end view on this drawing as being tangent to the main body housing but the side view suggests that the alignment is lower than the centre and therefore not tangent. The main body is tapered from the middle of the main body so a tangent with the lower struts would be difficult to achieve…I know I tried it and the geometry was complex.
I wondered if this should be tangential as is typical of these types of castings. Sometimes its not too clear what the intent of the designer was and this drawing sure would have benefited from the inclusion of an additional section through the strut.
I couldn’t be sure what the intent was here and the plan view did not present any new information, however I did notice a small detail for one of the side brackets with a solitary single line that turned out to be a reference for the interface with the struts and confirmed that in fact this item is not tangent!
This is the work-in-progress model which shows the generated profile of the junction between the main body and the lower struts, which satisfies the main details of the drawing.
As it turns out the strut is tangent but only to the edge of the end of the main housing and not the main body itself. There is also a glimpse of a side mounting bracket which will be quite complex to model due to the nature of the curved profiles that will need to be developed individually and then lofted to define the final surfaces.
Update: Finished model for the Tail Wheel Housing item 91-34003. To complete this sub assembly I just need to add the bushings.
These models and several others are now available on CGtrader
NAA P-51B/C/D Mustang: Radiator Coolant Mount
I discussed in my last post the development of a comprehensive drawing register for the P-51 and my rather ambitious intent to derive the list of parts associated with each sub assembly and main assemblies.
This could indeed be quite a task as for example on the P-51C alone we have 348 assemblies listed, some are sub assemblies and some are top level assemblies. The challenge is organizing the drawing parts list according to their assembly and retain the order of links on my filing system as per the main document register.
The NAA Numerical part lists (AN01-60JE-4 Section 2) give us some idea of how this data can be collated but the chart lists the top level assemblies and does not follow the hierarchy to the individual part. The individual parts though are listed in subsequent chapters of the parts list.
The part files themselves also contain information to assist with establishing the hierarchy of assembly; similar to the following for the Radiator Coolant Mount.
As you can see from the scans this part drawing typically lists the associated next level assembly, quantity and the aircraft variants to which they belong.
This image on the left is the next level assembly (sub assembly) which shows the inclusion of fittings and bushings and again lists another next level assembly.
Typically this is how the hierarchy works and its great that we can track the target assembly from the individual part drawings.
This is the top level assembly as noted in the above drawing. Our Coolant Radiator Mounts are highlighted in red.
In this example we don’t have all the drawings for the parts listed and though it would seem unlikely to be able to build this assembly with incomplete information it may be possible to interpolate sufficient data from what we know to develop the parts that are not available.
This is typical of these types of projects as the majority of scan drawing sets are incomplete and many parts can only be developed from physical examples or interpolated where we have the requisite data from other sources.
This approach is similar to how I plan to tackle this document register in identifying the links between the part files and the assemblies. We have the NAA register; which is a great starting point; and the part and assembly files themselves. There may be instances where the information from the drawings or the NAA register is unclear, in which case I would refer to other drawings in the series that may reference this information in the notes or comments.
At this stage I have transposed the NAA register assembly chart (noted above) into a spreadsheet format so that I can add additional key information.
The image shown here is a partial screenshot of how the fuselage data has been organised, showing the hierarchy level of the main assemblies according to their respective position in the NAA chart.
The first column is a reference number I use for hierarchical lists of this nature. There is still a lot of work to be done to collate the parts associated with these assemblies; hopefully most of which I will be able to transpose from the NAA scanned register.
In the interim I shall continue to develop some of these part drawings into accurate 3d cad models.
TechNote: MS Excel Pivot Tables: Drawing Register P-51
The drawing archive collection for the Mustang P-51 includes an NAA Document register in PDF that lists all the Scan Index Numbers, Drawings Number, Aircraft Type and Change (Revision)Number.
To make sense of this large archive; containing thousands of scanned images; it is necessary to first transpose the comprehensive NAA document register into a spreadsheet in order to analyse and filter the data according to requirements.
My requirements are simply to be able to group the data per content; Fuselage, Wings, Equipment etc; and per aircraft type; P-51A, P-51B, P-51C etc.
Further breakdown of data would involve isolating the main assemblies and then parts or sub assemblies belonging to each.
From Adobe Acrobat I extracted the pages of data as spreadsheet tables to which I added a Drawing Description and grouped the data sets together by “Content”…that took a long time to do as the extracted data first had to be checked and then sorted accordingly.
The drawing descriptions came from an index already created by Norman Meyers at Chanute Air Museum, so it was relatively easy to enter this data into my spreadsheet. Its a real pity I had not had access to Normans data earlier; could have saved me a lot of work. My thanks to Norman Meyers.
After sorting the data and inserting descriptions I now have separate worksheets for the content similar to this one.
What I really want now is to identify and organize the drawings belonging to each type of aircraft. For this exercise I use the Pivot Table function in Excel. Pivot Tables are great for organizing and summarizing data according to specific criteria.
Here I have initiated the Pivot Table function and selected the entire data-set of information relating to the Fuselage; as you can see we have a large number of drawings just for this one area!
When working with large data-sets it is good practice to select a new worksheet for inserting this new table.
What we end up with is a new worksheet with the pivot table outline on the left and a selection box on the right. We now select from the latter the columns of data we want…in this case all the main ones plus the P-51D; which will populate the outline table on the left.
Pivot tables by default include a summary row under each entry; I suspect this is more useful for statistics than organizing a document register; which we don’t want.
To remove the summary from the table we just need to select each column in turn using the small arrow as highlighted and turning off this option in “Field Settings” and select “None”.
The final step is to filter the data according to the required criteria; in this case I want all the drawings that have an “X” value in the P-51D column.
This is done by selecting this value from the header drop-down options; which lists by default the unique values in each column from the master table.
We now have a list of all the fuselage drawings and their location in the archive belonging to the P-51D aircraft.
The next step would be to extrapolate all the “assembly” drawings and from there the components that make up each assembly…but that’s for another day.
Pivot Tables are great for this type of job.
Many drawings of course are a shared resource for all variants. This drawing register has recently been updated with hyperlinks to all the drawings listed. See this post for details.
For further information on any of these projects please feel free to drop me a line via my contact page or email me at hughtechnotes@gmail.com
Messerschmitt: Bf109 Part
Scans of the drawings for the Bf109 have long been part of my archive, but I was always reluctant to spend anytime on this aircraft as many others have already done a great deal of work producing CAD data from these designs.
However and occasionally I come across an item that presents a challenge and this part is one of those.
8-10910-2501: Lower Frame mount.
This is a cast aluminium part with a 3 degree draft on all faces (refer second image showing cross section) except the lower edge and the top edge on centre.
The challenging aspect of this part has to do with the top face, where it was important to achieve tangency with the draft faces at each of the mounting holes and maintain this through to the centre portion where the face locally is perpendicular.
The material thickness also increases marginally at the extent of the 2 inward holes.
The front edge of this face has a center portion of a fixed radius; so to achieve the desired results based on the above criteria for the back edge I used a tangential spline. I could have just used a radius arc instead of a spline as the difference was only 0.2mm, but I am keen to get this stuff as accurate as possible.
It worked out rather well.
Spline Technote:
Splines are an absolute necessity when developing the finished profiles from the ordinate points; which occasionally throws up some unexpected results. Invariably at some point we need to manage the curvature of the splines in order to achieve the desired result.
The image shown here is a screenshot from the NAA P-51 fuselage station profiles, which shows clearly the ideal curvature for each station.
This image can be overlayed in CAD to serve as an aid to achieve the correct spline curvature.
Actually manipulating the curvature of a spline needs to be done in a manner that achieves symmetrical results on both sides of the fuselage station profile.
I was working on the tail-end profiles, which were giving me grief as the ordinates points were not sufficient to achieve anything close to the curvature I needed on the lower section.
In Inventor we have constraints for symmetry, which are normally applied when working on a sketch to ensure that changes on one side of a model are reflected exactly in the other.
Using this same technique I activated the spline curvature handles (A&B) on each of the points I wanted to be symmetrical (about center at C) and applied the constraint accordingly to the handles (red).
Now when I adjust one side of the curve the other side automatically reflects the changes.
I should note that the majority of curves generated from the ordinate points are usually very good; requiring very small if any adjustments; so its quite practical to spend some time in the areas where they are not so good.
At some stage the profiles will be lofted as a surface which would then be analysed to verify curvature and alignments.
NAA P-51D Mustang: Wing Geometry
Started work today on the wing geometry and ordinates.
This is the last ordinate data-set drawing I have in my P-51 archive and probably the most challenging, thus perhaps the reason why I left it until last!
The quality of the scanned drawing is not that great with much of the data missing or obscured and requiring a fair amount of interpolation to derive the correct values.
The interpolated data is derived through the use of various techniques within Excel, including polynomial curve formulas to determine the values I need from the known data.
So far this has worked out rather well enabling me to make a start on a geometry plan for the wing which will verify the relative dimensions of the Leading Edge, Front Spar and the 25% Chord line.
This drawing is still “work in progress”, which is shown for reference.
The wing ordinates are cross-referenced against 3 different sources to ensure correctness.
Update June 2018:
I have revisited this spreadsheet to include generated excel profiles to check the ordinates and also to derive the XYZ coordinates for input into CAD, centered about the front spar position. The missing and unknown values are now sorted thanks to a new resource…the spreadsheet is complete and verified. See Mustang Ordinates for full details.
North American P-51 Mustang: Air Scoop
Working with ordinates from these archive drawings can be a very time intensive operation. To give some idea of the content of this work I have just started working through the vast amounts of ordinate data for the Air Scoop and Oil Cooler.
This is a scrap view of the original NAA drawings showing the main ordinates for the Air Scoop.
This drawing shows 2 tables, one of which is the listing for the external contours and the other the internal contours.
The external ordinates comprises a total of 664 points and the internal ordinates comprise a total of 928 ordinate points.
Each point is manually entered into a spreadsheet which lists the Inch dimensions and then converted to Millimeter dimensions. The data has 3 values for the Station location, the Waterline (value along a horizontal axis relative to the ship ctr line at set intervals) and the Buttock line (value along a vertical axis relative to Frame Ref Line ).
These values are then processed using the concatenate function in Excel to extrapolate the required X,Y,Z coordinates.
The points are then grouped and imported into Autocad to derive a point cloud.
The first screenshot is all points combined with the local fuselage contours shown for reference; the second screenshot is the internal point cloud. All these points would then be contoured in Autocad to determine suspect locations and any orphaned points.
The external point cloud had 6 points prominently out of sync with everything else which turned out to be an error in the original data set. This is not uncommon and is usually quickly resolved.
Once I have an initial dataset that satisfies these primary requirements I would then import this data into Inventor or Solidworks for evaluation as a surface in each case.
At this stage, I have spent about 3 days on the data preparation and would expect to spend at least a week to properly evaluate the surface definitions.
It can be very satisfying work when you see for the first time all these data points translated into something tangible as a 3D model depicting the end product first realized all those decades ago.
Update: Decided to pull out all the stops and complete the datasets and point clouds:
North American P-51 Mustang: Fuselage
The drawing archive I have contains quite a large selection of legible fuselage frame drawings which I am collating according to the Station reference on the fuselage. I have a spreadsheet that lists all the Mustang drawings including the original drawing number, the scan image number and location within the archive.
Each fuselage frame at each of the designated stations may comprise 3 or more elements, which unfortunately are scattered throughout the many rolls of scans thus requiring some exhaustive work with the spreadsheet data-sets to sort the numbers and folder locations in order to identify and collate the required frames for each assembly per aircraft type.
One such frame was at station 216 which I decided to model; partly due to the fact I was getting fed up looking at and sorting spreadsheet data.
There are several methods to modelling this and whilst I was subject to the vagaries and still limitations of the Inventor product (Solidworks has more options for working with splines) I developed a workflow that obviates some of these limitations and also how the end product is finished.
One way of doing this is to simply create a surface for the main plane and then project a flange line along the edge to create a”folded” surface and then apply thickness but this method gave some unusual iterations in the smoothness of the fillets at the end of the profile. I found that the best way is to create surfaces for all six faces; the splines inside and outside, the top and bottom planes and the ends, then sculpt to create a solid.
I would then go out about creating the notches and cutouts in the solid and then shell the solid to the required thickness. This works very well and ensures the integrity of the original spline ordinate lines (which would have to be split to do this any other way). This method also maintains better continuity of the end fillets and curvature (image 2).
The frame drawings reference the mold line ordinates, which I have for the P-51 B/C Mustang variants.The P-51D is similar with the exception of the ridgeback on the main rear fuselage that has been reduced above the +10″ W.L.
Techy stuff: I mentioned a limitation in the Inventor software which relates to creating a line perpendicular to a spline. In Solidworks you just sketch the line and constrain it perpendicular to the spline, but you cant just do it this way in Inventor (as far as I know). What I did was use sketch construction lines to define the point of intersection with the spline that I wanted the perpendicular line to start from. As I already had a surface projected from the mold frame spline (for above construction) all I had to do was create a new plane perpendicular to this surface at the selected point. It was then quite simple to create a further sketch to define the line I wanted perpendicular to the spline at the correct location.
Aviation CAD TechNotes 