SBD Dauntless, from scratch

Ad: This forum contains affiliate links to products on Amazon and eBay. More information in Terms and rules

Wurger , Gnomey , thank you!

This relatively short post contains a digression about the aircraft shape. It was sparked by a suggestion that I received. Some time ago Alan from SOARING Simulator.com pointed me that the SBD NACA cowling was not as smooth as in my model (thanks, Alan!). He suggested that its contour was created from a combination of two or three arcs and a straight segment:


I thought about it and decided that this is a highly probable hypothesis. For most of the 20th century aircraft engineers did not have CAD systems. During that "BC" ("Before Computers" ) era the typical problem in the ship, aircraft, or car industry was: "how to precisely recreate in the workshops the shapes sketched — usually in scale — on the designers' drawing boards". The most important shape — the wing airfoil — was recreated using a "cloud" of data points. However, it was a time-consuming (i.e. costly) method. That's why for the less important areas, as the fuselage, designers used simpler solutions. The most obvious method to define a specific contour was a curve composed from two or three arc segments. It is relatively easy to recreate such a contour, because you need only to know the radii and the center point coordinates of the subsequent arcs. For example, there are many cases of such curves in the P-36 and P-40. There was also another drawing method for obtaining more "fancy" shapes (like the rudder contours) which was based on a general conic curves. To overcome this problem in a more advanced way the design team of the P-51 "Mustang" described all key contours of this aircraft using polynomial (2D) functions. Still the resulting points of the "Mustang" curves had to be calculated by hand!

The modern, computer-generated curves and surfaces (Bezier, NURBS, subdivision) have continuous curvature (as in figure "a", above). Thus it requires some effort to recreate in a computer model such a contour like the one sketched in figure "b", above), where the curvature continuity is broken between each segment. (BTW: the air flow "likes" the shapes that have continuous curvature. That's why designers always tried to preserve it in the airfoil contours).

All in all, I turned to the reference photos, trying to identify a kind of the contour like the one depicted in figure "b", above). Ultimately I discovered a more severe break than the lack of the continuous curvature: a minor difference in the tangent directions along the panel seam (i.e. the contour of this NACA cowling does not preserve even the tangent continuity!):


I marked the tangent directions along the panel seam in blue. This is a modern, high-resolution picture of a restored SBD-5. To exclude the possibility that this is an accidental inaccuracy made during restoration, I started to search for this break in all other photos. Surprisingly, I think that I was able to identify this "bulge" in the others SBD-5s. In the previous versions (SBD-1 to -4) it was hidden under the carburetor air scoop. But even there I think that I can trace it in the lines of the nearby panel seams (the gun troughs panels, side edges of the air scoop). Such a small deviations are usually a "side effects" of the technology applied to the particular element. Finally I used the reference photo to recreate this "bulge" in the side view:


What's interesting: previously the contour of this NACA cowling had a small convex break in the top view. (When I shaped it for the first time, without the additional section, I was not able to eliminate such a break in the tangent directions. It had to occur somewhere along this panel seam. I had only the choice where to place it, and I decided to leave it on the vertical contour). Now this contour is smooth, and there is a concave break in the side view.

I suppose that initially the forward ring of this NACA cowling was formed as a perfect solid of revolution. Then it was slightly deformed while fitting to the rear, "flat" part of the cowling. The cross section along the seam between these parts is not a perfect circle: it is somewhat higher than wider. Thus the rear edge of the forward cowling sheet had to follow this shape. It altered the tangent dimensions along this panel seam. In the top view it improved the fitting between these two panels of the NACA cowling. In the side view it only decreased the initial difference in the tangent directions.

Well, I hope that this post gives you a better insight, how we can deliberate on each small detail of the recreated airplane. In the overall picture of this aircraft the differences between the shapes before and after modification described above are hardly noticeable. However, I am a hobbyist, and sometimes we are the only ones who have the time to care about such minor things.

In this source *.blend file you can find this modified NACA cowling. (The change in its shape required some adjustments in the other panels).
 
I missed this thread. But reading through it I must say you show a lot of dedication and s skill. Being an open source user myself, it always amazes me how good many of this programmes actually are and you show us that you can do highly professional work with them. Great thread!
 
Wurger, Marcel, Gnomey - thank you!

In this post I will create the next section of the engine cowling. I copied its forward edge from the rear edge of the inner cowling panel. Then I extruded it toward the firewall:


I am going to split this object into individual panels, thus I already marked their future edges as "sharp" (as you can see in the figure above). It allowed me to preserve continuity of the tangent directions around these future panel borders from the very beginning.

In the next step I created the space necessary for the covers of the gun barrels:


Then I split this object into separate cowling panels:


I also used auxiliary "boxes" and the Boolean modifiers to cut out various openings in the side and bottom panel.

To keep the mesh topologies as simple as possible, I decided to model the inner part of the air outlet in the side cowling as a separate object:


(In the real SBD-3 it was also a separate piece of sheet metal). Its vertical contour was rounded to fit the fuselage behind the firewall (as you can see in Figure 48‑4), thus to shape it in this way I added three additional edge loops in the middle of this mesh.

The initial version of the gun cover was copied from the reference object, then I adjusted its shape fitting it to the adjacent panels (at least to their contours — see figure below):


Note that it is possible to have a corner in the middle of the border of a 3D surface that was carefully fit at its front and the rear contour (see the figure above)! In this case this is an intended effect, recreating the effect visible on the reference photos.

The last element of this cowling is the adjustable scoop (see the figure below), directing the air into the oil radiator (hidden inside engine compartment). It seems to have thick walls, but I suppose that they were empty inside (however, I am not sure — I cannot see any seams there):


I started forming this element by fitting its bottom surface into the fuselage contour:


Then I formed the side walls of this object. As the reference I used an auxiliary circle, centered at the scoop pivot point (as in figure "a", below):


Then I created the thick walls of this scoop using a Solidify modifier (see figure "b", above)

Initially I was going to round the edges of this object using a multi-segment Bevel modifier, placed after the Solid modifier. However it occurred that the Solid modifier created in some corners of this mesh dynamic faces that cause problems in the result generated by the Bevel modifier. Thus I had to "fix" the results of the Solid modifier before using the Bevel tool. You can see the rounded, thick edges of this scoop in figure "a", below, while figure "b" demonstrates the complete cowling panels:


Figure below shows the complete engine cowling, compared to an original aircraft:


Note that the model in the picture above uses different lighting than in the photo. It results in different shadows and reflections from the curved surfaces of the fuselage.

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

In the next post I will create the last panel of this fuselage: the hinged doors in the front of the windscreen.
 
Wurger, Gnomey - thank you!

In this post I will form the fuselage panels in the front of the windscreen. In the SBD there were two hinged cowlings, split in the middle. They allowed for quick and easy access to the M2 gun breeches and the internal cabling behind the instrument panels:


The parts of the fuselage around the cockpit are always tricky to model. It especially applies to the panel around the windscreen. When you obtain the intersection edge of these two objects, it can reveal every error in the windscreen or the fuselage shape. To be better prepared for this task, I created an auxiliary, simplified model of this fuselage section (see this post and the next one). Now I copied a part of it as the initial mesh of this panel:


In the next step I used my Intersection add-on to obtain the intersection edge between this mesh and the windscreen object (see figure "a", below):


Initially this edge is not connected to any of the mesh faces. To fit this panel to the windscreen I removed some of the original faces and created in their place the new ones. They incorporated the intersection edge into this mesh (see figure "b", above).

The resulting curve that I obtained in this way required just a few minor adjustments (see figure "a", below):


It is always good idea to check this shape in the reference photo (see figure "b", above). Fortunately, it seems that my edge between the windscreen and the fuselage fits its real counterpart.

When I verified the basic shape of this panel, I extruded it into the "frame" strip that spans around the windscreen:


To obtain a shape that resembles the real part, I assigned to the intersection edge a multi-segment Bevel modifier. It produces a fillet that forces the Subdivision Surface modifier (applied later) to generate a more regular rounding along this windscreen bottom frame.

Finally I created the armor plates that were attached to this hinged cowling. It was an easy part: I copied corresponding fragment of the cowling mesh, then I used a Solidify modifier to make it thick enough (on the photos it seems to have just a few millimeters):


I think that in this armor I will use textures (bump texture, ref texture) to recreate the bolts and the circular recesses around their heads (as visible in the photos). However, I will do it during the next stage of this project.

The last element that I modeled in this mesh was the seam along the bottom border of this panel. It was stamped in the sheet metal to overlap the upper longeron of the fuselage (see figure "a", below):


I also thought about recreating this detail in the textures, but ultimately I decided that it needs a more pronounced appearance. It is an easy effect (see figure "b", above) that required just a few additional edges (as in figure "c", above).

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

In the next post I will form the multiple segments of the SBD "greenhouse" cockpit canopy.
 
Wurger , Gnomey - thank you for following!

Like many contemporary designs, the SBD had a long, segmented ("greenhouse") cockpit canopy. In this post I will show you how I recreated it in my model. I will begin with pilot's canopy, then continue by creating the three next transparent segments.

I formed the pilot's canopy by extruding the windscreen rear edge (see figure below). (I formed this windscreen earlier, it is described in this post). The high-resolution reference photo was a significant help in precise determining its size and shape:


Generally, the canopy shape in the SBD is quite simple. The tricky part was that each of its segments slides into the previous one. (Oh, well, the pilot's canopy slides to the rear, but it does not matter in this case). This means that there were clearances between each pair of neighbor canopies that permitted such movements. If I made them too small or too wide, the last (fourth) canopy segment would not fit into cockpit rear border (i.e. the first tail bulkhead)! In such a case I would have to adjust back all the canopy segments. Well, I will do my best to avoid such error.

After the pilot's cockpit canopy I created the next, fixed segment:


It was a fixed part of the cockpit canopy, bolted to the fuselage. I used the available reference photos to precisely recreate its shape. Of course, I also had to determine the distance between the sliding pilot's canopy and this segment. It was a key moment: making it too narrow or too wide would spoil all further segments.

The photo references can be useful even when the modeled object is not visible: you can see such a case in figure below:


Although in the reference photo the gunner's canopy is hidden under the fixed segment, the last canopy element is in place. Thus I used its forward edge as the reference shape for the rear edge of the previous canopy. The front edge of this element is deduced from the cross section of the previous canopy segment, offset inside by the clearance distance.

Initially I created the last canopy segment by extruding such an "offset" rear edge of gunner's canopy:


The initial evaluation of the cockpit rear edge revealed that I had to extend a little the last edge of this object, to match the shape visible in the photos:


Of course, I also had to take care about the clearance between this and the previous canopy segment (see figure "a", below):


In the next step I cut out the unnecessary part of this mesh (see figure "b", above).

Finally I recreated the rounded corner of this segment. I did it using an additional vertex, located on the bottom edge of the last mesh face. (See figure below. In this way that face becomes an n-gon). When I slide this vertex forward along the bottom edge, it reduces the radius of this corner. A movement into opposite direction enlarges it:


Figure below shows the complete cockpit canopy:


Fortunately, its last segment fits well into the cockpit rear edge, so I do not have to adjust all these canopy segments! (I mentioned such a possibility at the beginning of this post).

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

The shapes you can see in red in figure above will become the transparent plexiglass surface. I still have to place the sheet metal frames on these elements (I will do it in the next post). There were also internal tubular structures that supported these canopies from inside. I will recreate them during the last, detailing phase of this project.
 
Wurger, Gnomey - thank you!

Today I will add the basic details of the cockpit canopy: its outer frames. However, before I started this work, I had to conduct yet another verification of the canopy shape. I placed the canopy rails on the cockpit sides, and verified if they fit the corresponding canopy segments. First I tested the rails of the pilot's canopy (see figure "a", below):


They were formed from open-profile beams (see figure "b", above). Why these rails are such an important test tool? Because they always have to be parallel to the fuselage centerline! It sounds obvious, but it can reveal various unexpected errors in the canopy shapes. In this case I discovered that the fixed segment of the cockpit canopy was mounted on the pilot's canopy rails. (In the previous post I assumed that this rail was placed between the pilot's canopy and this fixed canopy). If I did not find this error now, it would cost me much more work during the later stages of this project! Now I could quickly fix it (see figure "b", below).

In the rear part of the SBD cockpit you can find a double (two-beam) rail see figure "a", below). The forward segment of the gunner's canopy slides along the outer rail, while the rear (i.e. the last) segment — along the inner rail:


In figure "b", above, you can see that this rail protrudes from the last segment of the canopy. It's OK — in the real aircraft they cut out a half of its bottom edge, to make room for it (see figure "a", below):


Frankly speaking, these rails forced a lot of small modifications along this "canopy sequence". Their presence allowed me to fix various small differences between the reference photos and this model. In particular, now the roundings on the canopy rear edges match the photos, as well as the clearance between these canopies.

The typical cockpit canopy frame of a WW II airplane was a structure made from duralumin (or steel) tubes. In the SBD these tubes had rectangular cross sections, and were riveted to each other. They formed frames, which were covered with relatively thin (2-3mm) transparent organic glass plates. These plates were attached to the tubular "skeleton" by rows of small bolts:


The heads of these bolts had flat (conic) heads, which were "sunken" in the thin sheet metal strips placed over the organic glass plates. (There was also a seal layer under these thin duralumin strips). In this post I will recreate these external sheet metal elements. The internal tubular frame of the canopies will appear during the last, detailing stage.

As the first I created the windscreen frame (figure "a", below). The general method is always the same: I copied the mesh from the "glass" object, then cut out the frame stripes (figure "b", below):


Of course, the subdivision surface generated by these strips in some places does not lie on the reference "glass". Thus I had to adjust this mesh a little (as in figure "a", below):


Finally I obtained the result as in figure "b", above. Note that I created the frame of the hinged windscreen part as a separate object (just in case).

I formed the further canopy segments in a similar way. First I copied the mesh of the corresponding "glass" object. If it was required, I shifted it along its rails to match it against the reference photo (figure "a", below):


Then I inserted into this mesh additional, sharp (Crease = 1) edges along the borders of the frame strips that are visible on the reference photo. Finally I removed the unnecessary faces from the areas between these strips (figure "b", above).

When the frame shape matched the reference, I shifted it back onto the corresponding "glass object" (figure "a", below):


Finally I made this frame thick (by the "sheet-metal thickness" — about 0.02 or 0.03"). I did it using a Solidify modifier. It is directed outside, so it creates an illusion of thin stripes lying on the "glass" surface (figure "b", above).

In a similar way I created frames of the all other segments of this cockpit canopy:


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

As you can see, this Dauntless model starts to resemble the original aircraft. However, it still misses the propeller. I will work on it in the next post.
 
Wurger, Gnomey - thank you!

In this post I will form the propeller blades. The SBD Dauntless used two types of the Hamilton Standard propellers:
  • Hamilton Standard Constant Speed (counterweight propeller) used in the earlier Dauntless versions (SBD-1 … SBD-3). The blades of this propeller had smaller tips (see figure "a", below);
  • Hamilton Standard Hydromatic used in the later Dauntless versions (SBD-4 … SBD-6). The blades of this propeller had larger tips (see figure "b", below):

These two blades had different shapes. In this post I will recreate the earlier version, which was used in the SBD-1 .. -3 (figure "a", above). Several posts later I will modify its copy to obtain the later model of the blade, as used in the SBD-4 .. -6 (figure "b", below).

The main problem with recreating propeller blades of various historical aircraft is the lack of their precise drawings. In fact, I saw such a thing once, in the detailed plans of the Soviet WWII fighters. (Such a drawing contains the contour of the blade in the front and side view, as well as the set of subsequent airfoil sections, from the rotation axis to the tip). Nothing like this you can find for the typical Hamilton Standard blades! I looked for any trace of such drawings in the Internet. All what I found was a thread in one of the aviation forums. One of the participants of this discussion showed letters that he exchanged several years ago with the Hamilton Standard company. He asked for drawings of a blade that was designed in 1936. HS declined to reveal it, explaining that this design still remains their "business secret".

In such a situation, all what we have are the photos of the real blades. (Until somebody makes a 3D scan of such a blade, and will publish it in the Internet — I hope that such a reverse engineering is legal). I used these references to draw the most possible blade contour in my scale plans. However, I had to rely on the photos for best estimation of the size and thickness of the airfoil sections of this blade, as well as the variation of their pitch along the span (i.e. propeller radius). Thus they may be less precise copies of the original than the rest of this model.

Now I see that I should draw the propeller blade vertically or horizontally on my reference drawings (see figure below). Well, I sketched them at the fancy angle of 120⁰, but never mind — it is still possible to use it. I will have to slide and scale the mesh vertices along the local axes of the blade object.

I started the work on the blade by creating a cylinder object. Then I rotated it by 120⁰, aligning to the reference drawing (as in figure "a", below):


Then I extruded its upper edge and flattened it at the tip (as in figure "b", above).

In the next step I inserted a few additional edges in the middle of this blade (figure "a", below), and shaped them so they resemble a flat-bottom airfoil (figure "b", below):


You can find such a flat-bottom airfoils in the most of the propeller blades. Why? Because their flat bottom edge creates a kind of technological base in this twisted, complex shape. (For example, it allows you to measure the local pitch).

I do not know what was the airfoil used in the Hamilton Standard blades. In one of the aviation forums I have found that it was RAF-6. It is not confirmed information. If it would be true, the leading edge of this blade should be sharper (RAF-6 had smaller radius of the leading edge).

When the cross-section shape of the blade was set, I started to form its contour in the front view:


I stretched and shifted its airfoil edges until the blade contour fit the reference drawing. Figure "a", above, shows how the base (i.e. control) mesh of this contour looks like. In fact, I formed it directly, using the alternative display mode of the smooth resulting surface (as in figure "b", above).

Finally I closed this mesh along the circular tip. Comparing the reference drawing with the photos, I decided that the contour of this tip was a perfect arc. What's more, I decided that it was a little bit larger than on my reference drawings. Thus I created a reference object — a circle (figure "a", below):


I modified the last edge of this mesh, shaping the resulting contour around the reference circle. Figure "b", above, shows the final shape of the mesh, while figure "c" — the resulting tip surface.

Because I missed the information about pitch distribution along this blade, I decided to deform it in a dynamic way, using a curve via a Curve Deform modifier assigned to the blade object. In this case the deforming curve is a straight line, placed along the local Z axis of the blade (figure "a", below):


In such an arrangement I can control the pitch of this blade by changing the tilt value in the curve control points. At the tip the tilt is 0⁰, while at the last point (which lies on the propeller axis) it reaches the maximum value (25⁰). These values are just an estimation. The tilts in the middle points of the curve lie within this range (figure "b", above). It dynamically deforms the control mesh of this blade (figure "c", above). After a few trials I obtained the twisted shape that resembles the photos.


Figure below shows the three clones of this blade, arranged as in the propeller:


Their mesh is a copy of the original, with the "applied" (i.e. fixed) result of the Curve Deform modifier. Just in case, I preserved the original (not twisted) mesh of this blade together with the deformation curve in the References scene, among other auxiliary objects. It will be useful later, for the Hamilton Standard Hydromatic propeller, used in the SBD-5.

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

In the next post I will create the hub of this Hamilton Standard Constant Speed propeller.
 
Wurger, Gnomey - thank you for following . It's whole year!

In my post from the previous week I modeled the blade of Hamilton Standard Constant Speed propeller, which was used in the SBD-1, -2 and -3. The Douglas factory mounted on the hub of this propeller a small spinner (as in figure "a", below):


It seems that during the service of these aircraft, the ground crew often removed this spinner. It exposed the propeller pitch control mechanism (figure "b", above). There are many photos of the SBD-2 and SBD-3 without spinners, thus I decided that I had also to model this "bare" variant.

These constant speed propellers were in wide use during the 30', but it was not easy to find any detailed photos or sketches of their counterweight pitch control mechanism. Finally I figured out that the counterweights were connected to the corresponding blades (figure "c", above). The central cylinder shifted along the propeller axis, controlling the angle of the counterweight arm (and, in the effect — the angle of the propeller blade pitch).


How to start forming such a complex shape like this variable pitch mechanism? I began by identifying its key axes and base planes:


The most difficult part of this process is not visible here: I had to realize the basic shape of these parts. I spent some hours studying the photos before I decided that the hub (referred also as "barrel") of this propeller can be composed from several cylindrical elements. After this conclusion I could start forming this object. I began by shaping a cylinder around the blade base:


To facilitate my modeling, I used all of the symmetries that exist in this part. I formed just a quarter of the cylinder mesh, then mirrored it across the blade axis. In the next step I placed clones of this mesh around the two other blades. (When I modify the original mesh, it will also modify these clones).

I formed the side surfaces of this barrel from a half of an elliptic cylinder:


I started it as a classic cylinder of a circular base. Then I rotated this object by 60⁰ and scaled it along its local Z axis until I obtained the shape resembling the photos. Then I used my Interesction add-on to obtain the intersection edge between this surface and the neighbor cylinder.

Finally I joined these two meshes and removed all of their faces. I preserved just the three edges: around the blade base, the inner edge of the elliptic cylinder and the intersection edge. I joined them with the new faces (figure "a", below):


In this way I obtained a solid which looks like the original propeller hub. Note that the three segments "touch" each other without visible seams — it looks like a continuous surface (figure "b", above). I used a multi-segment Bevel modifier to generate regular fillets along the sharp edges of this mesh.

I could make the opening for the counterweight arms in the front of this barrel using a Boolean modifier. However, it was relatively easy to recreate this particular shape by small modifications in the control mesh (figure "a", below):


In a similar way I integrated the small fragment of the cylindrical rear axis with the rear faces of the barrel mesh (figure "b", above).

The next difficult element are the flanges for the bolts that in the real propeller joined the two halves of this barrel. If I tried to incorporate them into the barrel object, it would significantly complicate its mesh. In the effect I would spent some additional hours on various adjustments. Thus I decided to create this flange as a separate object, that joins with the main body of the barrel in a more-or-less seamless way.

Figure below shows how I shaped this element:


As in the case of the main body, I model here just a single segment of this flange — the rest is replicated by the Mirror modifier. I started from a cylinder created around the eventual bolt axis (figure "a", above). Then I modified this mesh, fitting it to the underlying surface of the central barrel (figure "b", above). In the next step I extruded additional "strips" of the faces that lie on the barrel surface (figure "c", above). I rounded the sharp edges around these faces with a fillet (generated by a multi-segment Bevel modifier). Finally I extruded the part of the small wall that accompanies these bolt flanges (figure "d", above), and shifted its origin point onto the propeller axis, to create similar flange on the opposite side of the blade base.

I created two additional clones of this object, placing them around the blade bases (figure "a", below):


I also made the walls of this barrel thick, using a Solid modifier. Looking at figure "b" (above) you can see that these bolt flanges look now like integral parts of the barrel. I will recreate the remaining elements of this hub during the detailing phase (for example — the bolts that kept halves of this hub together).

The last element I have to model now are the counterweight bearing shafts. They were attached to the control cylinder. I started this part by adding a small shaft bushing on the bottom side of the counterweight arm (figure "a", below):


Then I created the first row of vertical faces around this cylinder (figure "b", above). I also created two additional clones of this mesh and placed them around the control cylinder, below corresponding counterweight arms. When they fit each other, I modified the topology of these faces, preparing them for joining the mesh of the shaft bush (figure "c", above). Note that I placed these faces precisely below the second-last pair of the cylinder vertices. I also prepared additional faces, which will allow me to quickly join them with the octagonal cylinder of the shaft. Finally I duplicated these vertical faces, placing them on the opposite side of the shaft mesh. It allowed me to quickly create the last missing faces in this mesh, and obtaining the finished shaft bushing object (figure "d", above).

Figure "a" (below) shows the complete pitch control mechanism of the constant speed counterweight propeller:


I will add more of its details (the bolts, for example) during the last phase of this project. In the last step I also recreated the spinner (figure "b", above). It was an easy solid of revolution — I will not elaborate here how to create it. (If you want to learn more about shaping the spinners — see this guide). There is one missing thing: I do not know how this spinner was attached to the propeller hub. Do you have any hint on this subject?

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

I will create the R-1820 engine during the detailing phase. This was the last element of the SBD-3 that I created before the UV-mapping and texturing phase. (In fact, I am not going to unwrap the small elements of the variable pitch mechanism. I could just create the propeller blades and the spinner. I created it just because this propeller hub is a quite large part of the Dauntless silhouette). In next post I will recreate the few details which differ the SBD-1 from the SBD-3.
 

Users who are viewing this thread