Injection Molded Rowing Machine Handle (summer 2020)
The existing handle for the hydrow rowing machine had a high failure rate in the field. Some percentage of handles would snap in half in users’ hands. Upon inspection of returned broken handles, our department noted cracks near one of two welded joints on the handle. We attributed this to the poor heat contol in a hand-welding process used at our CM to attach two small tabs in the underlying structure of the handle. If an operator held their welding gun too close to the part, dwelled in one area for too long, or used too high an amperage on their welder, the area surrounding the weld would harden and become brittle too much for our application.
An injection molded design offers a higher degree of consistency between parts than a design which requires hand welding. In our case, it also had a potential to reduce per-part costs, because of the high manufacturing volumes.
Part Breakdown
I knew the outside shape of the handle was to remain exactly the same as the original handle. So, I began this design by re-modeling the original handle shape using features that I could understand and work with (the original handle had been designed by our CM).
Next, I ideated on an optimal breakdown for manufacturing by injection molding. I knew I had to break the handle in half, approximately at its centerline, to avoid undercuts. I also saw no reason to change the overmolded grip design, so I knew that if the handle was to be separated in half at its centerline, areas of the handle beneath the overmolded grips would have to be assembled before overmolding, while access to the center area would be required for attachment of the handle to the rowing machine.
I split the main structure of the handle in to four parts - ‘Handle Base,’ and three cap parts ‘Right End Cap’ ‘Left End Cap’ and ‘Mid Cap.’ I used three cap parts so that two caps could be installed before overmodling the EPDM grips, and one cap could be removed and installed separately in order to install the handle to the rower. This three-cap breakdown also enabled me to move the mating line between the handle base and end caps further toward the bottom of the handle from the mid plane of the grip area, increasing rigidity.
Ribbing / StructureNow with an empty handle shell surface modelled in solidworks, I had to define a ribbing structure for the handle. I initially split the ribbing in to three groups
I extruded ribs through the handle, and cut them at their junction to the handle using the driving planes I created for the handle. I defined the ribs’ maximum protrusion from the handle using a set of surfaces I made especially for the ribbing.
I determined the exact layout of these three rib groups by running FEA on each iteration, and tweaking the rib locations and heights to spread stress. I do not have record of how many iterations were required to reach my final design, but I believe it was in the region of ten.
I also added structural ribbing to the cover pieces. In the below image at left, you can that the ribs in the left and right end caps are designed to engage the grip ribs in the handle base and increase overall rigidity. The center image below shows in the mid cap, where I some ribs ran along the surface to prevent warping during cooling out of the tool. The right-most image shows a small reveal between the center cap part and the handle base. This allows the screws and three bosses to define the relative locations of the handle base and center cap, without over-constraint from the outer rim of the center cap.
Early Prototying I sent my design out for 3D printing. This allowed management to see my design in 3D, and for me to assess my execution of the interfaces between parts. It is hard to assess fitment exactly from 3D printed models, as dimensional control isn’t perfect. However, it was helpful for me to get a sense of how the overall design would come together.
Reconsidering Interface to the Rowing MachineThe rowing machine is driven by a long piece of 1” wide polyester webbing (looks like a skinny seatbelt). This webbing wraps around a spool inside the machine, and rotation of that spool actuates the resistance mechanism inside the machine. The webbing runs out of the machine and in to the handle.
The original handle, which I was replacing, retained a loop at the end of the webbing using a standoff fixed in place by an m6 thru-bolt. This thru-bolt interface concentrates pull forces in a small area, which is not optimal for nylon plastic. To spread stresses over a larger area, I considered several fittings to attach the webbing to the handle
D Ring:
Originally provided to me as the agreed-upon drive webbing interface, this option spread force over a large area, and has a very high strength. However, this option requires permanent joining of the d-ring and drive webbing. This necessitates replacement of the machine’s drive webbing in retrofitting old machines to this design, and some more complexity to subassembly processes at our CM. I ran FEA on this option in a mild steel, as a baseline for comparison of other concepts:
D Ring with Rear Split:
This iteration has a slit on the rower side (“rear”) of the handle, through which the webbing is installed. With the same fixtures and applied forces as the solid d-ring, FEA in mild steel suggests 21% higher stresses than the d-ring with no split.
D Ring with Front Split
This iteration moved the slit to the drive webbing side (“front”) of the handle. FEA in mild steel shows a decrease in maximum stresses in this iteraiton when compared to the rear split verision. However, the added gap in the interface between the ring and drive webbing could damage the drive webbing, or allow the webbing to pass out of the d-ring.
Square Ring with Rear Split
In an effort to limit the forces working to spread the d-ring to either side of the slit, I modelled a square version of the d-ring.
FEA showed that despite contact faces perpendicular to rowing forces, this ring has much less structural integrity than other options.
Clevis D-Ring
This clevis model creates a closed loop with no slit, which creates structure that the slit versions lacked. This version uses an M4 bolt to retain the drive webbing, allowing removal and installation of webbing.
FEA shows maximum stresses of 2.3e+8 N/m², suggesting similar structural integrity to the rear split d-ring. The cross section of the M4 bolt is smaller than the that of the prior d-rings, so despite the closed loop the M4 bolt section exhibits stresses similar to those seen in split rings with a larger cross section.
Winged U fitting
In the end, I presented this FEA to our management, and they told me to design a fitting optimal for grounding in to the handle and webbing installation. They told me we would simply use a stronger steel alloy to make up for any weaknesses, within reason. The stronger steel alloy added a very marginal cost on this small part.
Handle Model Rework and Final ChangesIntegrating the winged U fitting in to the design required a rework of the center ribs and shoulder ribs. I passed this design off to our CM, who requested two additional features for grip retention on to the handle.
Our CM requested the thru-hole feature shown above, which simply required the addition and integration of large bosses in the handle base and left/right end cap parts. Our CM also requested a channel at the grip’s edge, which required lots of iteration, and ultimately a wall thickness inrease of the entire part. I did not make this part fully parametric, so changing the wall thickness required remodelling of the driving surfaces of the part, and some subsequent editing of the rib definitions.
First Shots to Design Validation
At this stage, our management approved the design, and ordered tooling for the parts. Below is an image of one of the first shots.
Our validation engineer worked with our CM to design a test fixure and test process. The handle sustained almost four-thousand Newtons of load in a max load test, and a sample sustained eleven million cycles on an accellerated life testing rig. A handle designed by an outside firm had sustained fewer than fifty thousand of the same cycles, so eleven million was a great result.
I worked with our quality engineer to review and refine the SOP at our CM, to ensure all critical assembly steps were called out.
In 2022 we replaced the initial production handles with this new handle design and sold approximately fifty thousand units. To my knowlege no handles ever broke during normal usage, although a few did fail due to shipping damage (200 lb rowing machine falling on the handle). The handle is still in production today at hydrow, and hasn’t required any changes since its release in 2021.