SBD Dauntless, from scratch

Looks great so far!

Wurger, Gnomey - thank you for following this thread!

After the previous post I decided to simplify the empennage fairing. Originally I created it from two separate objects: the fin fairing and the tailplane fairing, split across their fillet. Now I decided to eliminate this troublesome seam by joining these two meshes into single object:



I will split it later, along the bottom rib of the fin (there was another panel seam in the real airplane). To simplify creation of the original overlapped panels, I simultaneously split the fin into the forward and the rear part, along one of the original seams.

As you can see in figure above, there are the same number of spanwise edges on both fairing meshes. It is a matter of sheer luck, but it makes the process of joining these two parts much easier.

First, I modified these edges, bringing them closer to each other. (I did it by sliding their vertices along perpendicular edges — as you can see in figure "a", below):



Then I removed the unnecessary faces (as in figure "b", above).

Finally I filled this gap with new faces, effectively merging these two meshes (figure "a", below):



In the side view (as in figure "b", above) you can see that the forward part of the fin extends a little the panel line visible on the reference drawing. It will be overlapped by the forward part, which will end precisely along the original seam. (Such an overlap is visible on the photos). Note also that the seam along the fuselage upper panel (marked on the drawing by the dotted line) is somewhat higher than on my scale plans. This is the effect of the modification that I made in the upper part of the fuselage (described in my previous post). My reference drawings are simply wrong about its location in the side view.

I think that the empennage fairing looks much better after this modification:



It fits well all the three elements it joins: the fuselage, the fin and the tailplane. The fillets looks smooth and natural. As you can see, I split the fin and the fairing along the bottom rib edge, as I planned. (There was original panel seam). I think that this new arrangement of the model objects will facilitate further detailing of this assembly (for example, now the forward fin panel overlaps the other elements, as in the real airplane).

The last element of this assembly is the fin tip: in the real airplane it was stamped in one of the fuselage inspection doors. I started to form this part by creating a plain, rectangular cover placed over the fuselage, and separating the corresponding fragment of the fin tip (figure "a", below):



Then I joined these two object into a single mesh (figure "b", above).

In the next step I adjusted corresponding edges of both elements, and removed the unnecessary faces (figure "a", below):



Finally I filled this gap with new faces. Finally, after some rearrangements of the mesh topology, the resulting elements looks like in figure "b", above).

Figure below shows the final object, fitted to the fin and the fuselage:



The last element that I need to finish in this empennage is the rudder leading edge. I created it in the same way as the leading edge of the elevator: from a single circle (see my post from January 9th). I extruded it into a cone (figure "a", below), then removed the unnecessary faces and created new ones (figure "b", below):



Figure below shows the completed empennage (note that I also created the fin spar):



The last missing element is the tail tip. It has a rather complex shape, so I started modeling this part by copying its outer edges from the fuselage, fairing, rudder and elevator. You can see them in the picture above. I do so when I have no clear idea how to start. In the next post I will describe what I did next.

In this source *.blend file you can evaluate yourself the model from this post.

Good work so far!

Very nice! My worry is, with everything going digital these days,is physical modeling going to go the way of print newspapers?

Cheers,

Gerry

Wurger, Gnomey, Gerry, thank you!

Gerry said:

(...) My worry is, with everything going digital these days,is physical modeling going to go the way of print newspapers?

Click to expand...

I do not think so. I think that all these branches of the modeling will coexists. For example: the paper models were the first ones (they appeared at the beginning of 20th century). In the middle of the 20th century the plastic models were introduced. They did not wipe out the paper models. Now I we have yet another branch: digital models. As you can see, they require a lot of work, if you take them seriously. On the other hand, "plastic" or "paper" modelers can use them as an additional reference (as you can see in this thread, they pass much harder verification than any scale plan). Our goal remains the same: recreate the historical airplane as precisely as possible.

The tip of the SBD tail was a light fairing, attached to the last bulkhead (at station 271 — see figure "b" below). That's why you can see "NO PUSH" label on the photo in figure "a". The tail wheel was attached to the bulkhead 271, which transferred the resulting loads forward, via the tail structure. The tail tip fairing was always free of any significant loads. However, the shape of this part is a combination of the empennage fairing and the last fuselage segment. What's worse, there is a large opening at the bottom — for the eventual tail wheel deflection:



I had no initial idea about the mesh topology that I should use for such a part. Thus I started by copying all of its external edges from the adjacent objects (see figure "a" below):



Then I worked out an idea of its topology by sketching possible mesh edges on a scrap of paper. (I often do that before I start modeling a complex mesh). You can see the scan of my sketch of this tip in figure "b", above. I "think by drawing", so this method helps me to better realize the shape that I have to create. These working sketches object do not have to perfect. The more important thing is the order of the individual edges and vertices (identified by the numerical IDs).

When the mesh topology on the sketch looked simple enough, I started building this mesh by extruding the trailing edges of the elevator fairing (see figure "a" below):



Then I extruded the upper edge of the elevator fairing into another rectangular "patch" (see figure "b", above). In the next step I extruded similar "patch" from the rudder contour (see figure "a", below):



Finally I filled the gaps between these three mesh "patches" by creating two rows of new faces (as in figure "b", above).

The resulting subdivision surface required some minor adjustments of the control mesh vertices. They formed the proper shape of the mesh behind the elevator trailing edge (figure "a", below):



While sketching this tip, I decided that it will be better to close it with a separate object — the "closing strip". Such a strip reproduces the original piece of the sheet metal that contained the frame of the running light. (I thought that the side-view contour of this tail tip might require different edge distribution than the mesh of its sides). At this moment I created the initial segment of the "closing strip": the part around the running light frame. Then I used it as the reference object for shaping the mesh of the tip sides (see figure "b", above)

I split the model surface into separate objects when I expect significant differences in their mesh topology. I "mask" outer edges of such objects by placing them along the original panel seams.

Click to expand...

When the upper part of the tip was ready, I started forming its bottom part. As you can see in figure below, I did it in the same way as the upper fairing. First I extruded a part of the bulkhead contour (figure "a"), then I created new faces to incorporate this patch into the main mesh (figure "b"):



I extruded the side faces of this tip from the bulkhead edge (figure "a", below), then filled the gap in the resulting mesh by creating a row of new faces (figure "b", below):



(Note that I had to create more "bulkhead" edges in this mesh than I originally sketched (see the second figure in this post). Some of these edges came from the original vertices of the elevator edge, while the other were required by the shape of the bottom edge (around the tailwheel opening).

The ultimate number of edges in a mesh is often the sum of the vertices required to obtain appropriate shape on its opposite border edges.

Click to expand...

Finally I extruded and merged the last part of this tip:



As you can see in figure "b" above, I added an additional, diagonal edge below the trailing edge. I did it to obtain a better shape of the elevator fairing

Figure below shows the smooth resulting shape:



I had no photo taken from the top or the bottom that would precisely reveal the vertical contours of this part. Thus I assumed that this tip is a smooth continuation of the tail cone (as I marked with the dashed line in Figure 38‑9). In the next post I will verify this assumption using available photos.

In this source *.blend file you can evaluate yourself the model from this post.

Lovely work so far!

Wurger, Gnomey - thank you!

In the previous post I formed the shape of the SBD Dauntless tail tip. In this post I will finish its "closing strip" that contains the running light frame. I will also verify the overall shape of the tail tip using the available photos.

There is one thing I didn't mention in the previous post, just to keep the narration focused on the pure modeling. Before the modeling I carefully studied the reference photos. In the result I found differences in the shape of the curved trailing edges of the fairing behind the elevator. On the photos you can see a straight fragment of this edge (figure "a", below). Its presence means that the curve of the trailing edge was smaller, and the fuselage was somewhat thinner here. You can see the differences between the real shape and my reference drawing in figure "b", below:



I did not notice these detail before. As you can see, I applied this modification when I started to model this part.
When the mesh of the tail tip was formed, I worked on the "closing" strip. I created a part of it as a separate object in my post from 2016-02-13. Now I extruded it along the side contour (see figure "a", below), then extruded the side faces of this strip (figure "b", below):



I rounded the sharp edges of this element with a fillet. It is dynamically generated by a multiple-segment Bevel modifier — you can see the result in figure "a", below:



However, when I compared the bottom part of this tip to the photos, I saw that there are significant differences! Jut compare it in figure "a" and "b" above. The side edges of the tailwheel opening are less curved, and its rear edge is wider.

What is the reason of these differences? So far I tried to shape the bottom part of this tip as the smooth continuation of the previous tail segment (see figures "b" and "c", below). It seems that I was wrong: these lines were broken at station 271, where the tip fairing was attached to the last bulkhead of the tail (see figure "a", below):



To correct this shape, I first made the tailwheel opening wider by rotating the bottom part of the mesh:



Finally I adjusted the shape of the "closing" strip to this new opening (figure "b", above). Now it resembles the original in the photo.

When I fixed the shape of this opening, I noticed another difference, this time in the shape of the tip cross-section. It is revealed by the vertical panel seam behind the tailplane:



This vertical seam seems to be flat, especially in the restored SBD from figure "b", above. In my model this line is much more convex (figure "c", above).

The primary reason of all these differences between my model and the real airplane is the lack of the reference: I have no photos of this fuselage tip taken from above. Thus I have to determine its shape on the plans by various indirect means — and assumptions. In such cases, when you shape it as a 3D model, you will often find errors in the reference drawing.

Click to expand...

Well, to make this section more "rectangular", I have to make the fuselage even thinner in this area:



Yet the photos reveal another small difference, this time between the two different restorations of the SBD Dauntless: compare the photos "a" and "b" in the second-last figure above. The aircraft from photo "a" has the edges of the tailwheel opening bent inside, while in the SBD from photo "b" the tip cross-section contour is straight to the end. Which one is true? At this moment I do not know! However, it will be better to modify the mesh of this tip in a way that makes such a rounding possible. (You can always straighten a curved surface. However, bending a flat mesh requires additional edges). That's why I modified the mesh topology around this opening, adding another "longeron" edge (figure "a", below):



Then I shifted two bottom edges a little (figure "b", above), forming such a cross section as you can see in photo "a" (compare it with the shape in figure "c", above).

Below you can see tip, finished for now:



I will recreate its internal reinforcements (bulkheads, stringers) later, during the detailing phase. At this moment I do not know whether I have to modify this part in the future. (It may happen when I find a better reference materials). Such a modification would require adjusting these internal structures, so it is better to postpone their creation as long as possible.

In this source *.blend file you can evaluate yourself the model from this post.

In the next post I will start to work on the engine cowling.

Good work so far!

Great work Witold. I was reading through your earlier posts concerning the different lengths given for the various SBD models and decided to peruse some of the documentation I have on hand. In the pilot manuals for the SBD-3 and SBD-4 a length of 31' 8.75" is given, a full foot shorter than what you found. Any thoughts on the discrepancy?

Interesting! Do these manuals specify if this is the "level length" (i.e. measured along the fuselage centerline), or the "length on wheels" (i.e. measured along the ground level)? In my posts I discussed the "level lengths". The corresponding "length on wheels" is shorter: see the dimensions on this SBD-5 drawing: while the "length level" was 33' 1/8", the "length on wheels" was 6" shorter: 32' 6". Thus we can deduce that for SBD-3 the length on wheels was: 32' 8" - 6" = 32' 2".

However, even if this 31' 8.75" was the "length on wheels" (more practical dimension in the case of airplane handling), I cannot explain the remaining difference of 6".

For the scale plans and the 3D model presented in this thread I used a "semi-empirical" approach: I relied on the explicit dimensions from the Douglas general arrangement diagram. The geometry of the SBD airframe behind the firewall was the same in all versions [an assumption, but of high probability, based on the all available photos]. You can find on these Douglas diagrams the key airframe dimensions - the firewall station (0), as well as the locations and basic "trapezes" of the wing, fin, rudder and tailplane. I used them to determine the position of my model on the reference photos. Then I copied the shape of the engine cowling as it appeared on these photos (without any previous assumption about the overall length). In the result I obtained for the SBD-3 & 2 the level length of about 32' 8", which agrees with some of the available sources.

The length is listed as "Length, Over-All." No mention of "level length" or "length on wheels."

Well, I have looked again into another source document: data from SBD-3 performance tests, made by Bureau of Aeronautics in 1942. I discovered that its page 2 specifies both: the level length and the length on wheels (it seems that I was wrong in my estimation by 2" - it was 32' 4").
As you can see none matches to the length from the Pilot's Manual. That's why I used the "semi-empirical" method that I described in the previous post.

Back to modeling: I started working on the engine cowling:

SBD Dauntless had a radial engine hidden under typical NACA cowling. The Douglas designers placed its carburetor air intake on the top of this cowling, and the two Browning M2 guns behind it. In the result, the upper part of the SBD fuselage, up to the pilot's windscreen, had a quite complex shape:


I am sure that I will tweak this shape multiple times before I reach the most probable compromise between all the reference photos I have. It will be much easier to do it by modifying a simple mesh instead of the complex topologies of the final cowling. Thus I decided to create first a simpler version of this fuselage section and adjust it to the all of the available photos. I will describe this process in this and the next post. Once this shape "stabilizes", I will use it as the 3D reference in forming the ultimate cowling. Because I am going to recreate all the internal details of the engine compartment, I will create each cowling panel as a separate object.

I started by creating the three key contours of the NACA cowling:


Section 1 (see figure "b", above) was a perfect circle, while section 2 was a little bit higher than wider (see figure "a", above). Ultimately section 3 was a regular ellipse. Note that all these sections have the same number of the vertices (32).

Once I created these three edges, I connected them using three arrays of faces. Them I added in between (using the Loop Cut command) three additional edge loops. I needed them to form the curved forward part of this cowling:


As you can see above, I fit the silhouette of this NACA cowling to the reference photos. This is a photo of the SBD-5, so I moved this cowling forward by 3.5" (see in this post about differences between SBD-3 and SBD-5 cowlings Figure 4-6, for the explanation).

When I finished the NACA cowling, I formed the basic shape of the next fuselage section:


I created this part in the same way as the previous one. First I copied and shrunk the last NACA cowling edge, creating the gap for the outgoing air. Then I copied the firewall edge, and joined these two edges by an array of new faces.

In the next step I extruded the upper part of this surface, creating the section below the windscreen:


It is a "linear" continuation of the previous fuselage segment. I fitted its sides to the mid-fuselage, which I formed some months ago.
The next elements are the "bulges" that covered breeches of the Browning M2 guns. In this 'quick and dirty' approximation I formed them from a separate mesh patch (see figure "c", below):


Of course, I verified their shape on the available photos, as you can see in figure "c", above.

This comparison revealed, that the intersection lines between these "bulges" and the main fuselage require some improvements: they have to resemble straight lines:


To obtain such an effect, I had to decrease the upper radius of the last bulkhead (figure "b", above). In such a simple mesh it required just to move a few vertices. If I had to perform such an operation on the final panels, it would be much more difficult!

In the front of the gun barrels there were long recesses in the NACA cowling. The outer edges of such a feature are always a tough test for the model, because they depend on the proper shape of the both intersecting elements. First of these objects is the NACA cowling, the second is the shape of this recess — a cylinder in this case:


I cut these openings using a Boolean modifier. Figure below shows the result:


In figures "b" and "c", above, you can see the evaluation of the obtained contours. They seem to fit the borders of the recesses on the reference photos. (There are minor differences, but I suppose that they are results of the rounded edges of these features.

In this source *.blend file you can evaluate yourself the model from this post.

In the next post I will continue shaping this first approximation of the engine cowling.

Nice work so far!

Wurger, Gnomey, thank you!

In this post I will continue my work on the engine cowling. I started it in the previous week by forming a "first approximation" of the forward part of the SBD Dauntless fuselage. Now I will create the last elements of this auxiliary object.

First of them are the covers around the M2 gun barrels. They were hinged around their inner edges, and their cross-section varies from a semi-circle at the NACA cowling to a flat line at the firewall:


I started forming this cover from a conic cylinder, created around the gun barrel:


Then I cut out its bottom part and flattened its end section along the side "bulge":


I formed it to resemble the gun barrel covers as they were in the SBD-1..-4s. Studying the photos I identified that this detail looks a little bit different in the SBD-5 and -6:


As the last element of this auxiliary object I will form the windscreen. I need it for determining the ultimate slope of the "bulges" around the breeches of the M2 guns, and for checking the shape of its intersection with the fuselage. (If I did it later, it could reveal some unexpected surprises about the fuselage geometry, resulting in additional work).

I used the reference photos to determine the basic radii of the canopy hood and the windscreen cross sections:


(In this aircraft canopy hood slide under the windscreen, thus the radius of the windscreen cross-section was a little bit larger). As you can see in Figure "b", abve, the obtained contours differs a little to my reference drawings. (It seems that on these drawings the top of the cockpit canopy is a little bit lower than I have ultimately found it now on the photos).

In the next step I determined the radii of the cylindrical fragment of the windscreen:


It seems that it was not a regular cylinder — its radius at the top of the windscreen seems to be larger than the radius at the bottom (figure "a", above).

I created this cylinder as the first part of the windscreen surface (figure "a", below). I verified its shape using another reference photo (figure "b", below):


In the next step I removed the rear part of this cylinder and formed the flat, triangular side plates of the windscreen. As you can see in figure "b" above), they were hinged, providing the maintenance access to the M2 guns on the cockpit sides.

Then I extruded two additional rows of faces, forming the upper part of the windscreen (figure "a", below):


When the shape of the intersection between the windscreen and fuselage matched the reference photos, I also verified its side contour (figure "b", above).

Figure "a" below shows the complete object that approximates the shape of the SBD engine cowling. I set its color to red, as I do for all the reference objects in this model:


In figure "b" above you can see that it fits pretty well the reference photos. This is a picture of the SBD-5 from Chino Air Museum. The SBD-5 and -6 had their engines and the NACA cowlings shifted forward by about 4", so did I in this model (see in this post Figure 4-6 for details).

In this source *.blend file you can evaluate yourself the model from this post.