Rev Up Outboard Production
Rethinking product and process allowed Mercury Marine to rev up outboard production
By Jim Destefani
Mercury Marine has been an innovator in marine propulsion systems since its founder developed his first outboard motor more than 65 years ago. Headquartered in Fond du Lac, WI, the company today has manufacturing facilities in Fond du Lac; Stillwater, OK; St. Cloud, FL; Juarez, Mexico; Petit-Rechain, Belgium; and Newton Abbot, UK.
In Fond du Lac, 1,000,000 ft2 (93,000 m2) of manufacturing space in Mercury's machining and assembly plant is devoted to production of outboard motors in a wide range of horsepowers and other characteristics. Other facilities at the site include a die-cast and lost-foam casting plant as well as propeller manufacturing.
Recently, Mercury launched its Verado line of supercharged four-stroke outboard engines. The only supercharged four-strokes in the outboard market, Verado engines are available in 200, 225, 250, and 275-hp (150, 170, 185, and 205-kW) six-cylinder versions as well as 135, 150, and 175-hp (100, 110, and 130-kW) in-line four-cylinder models.
The Verado engines (see sidebar for details) have raised the bar in outboard engine performance, according to Mercury Marine's Mike Oswald. And, he says, the company felt that such a revolutionary product should be manufactured on a new line, employing automation where applicable as well as lean manufacturing concepts.
The project manager for the line, Oswald has since transitioned into the role of lean/six sigma black belt. He says the development of the Verado assembly line project was unlike anything Mercury had previously done.
"We approached this development process completely different than in the past," Oswald recalls. "We had program managers for engineering [design] and manufacturing, and we had groups who worked with those managers. Early on, we asked the collective bargaining unit to have some production employees taken off their regular responsibilities to become part of this team. In the beginning, we had three people from production in addition to our engineering, maintenance, and other support groups. So our engineering technicians and production employees built the prototypes together."
Oswald says the cooperation progressed through assembly system design and into prototype, design validation, and production validation builds. "We added team members as needed to help with ergonomic assessments of the line, and other tasks," he says.
The new approach extended to assembly technology supplier selection as well. "Traditionally, we would have designed our assembly processes, then gone out and found a medium-size machine builder or conveyor company to work with us. We would have been the main project leader and integrator," Oswald explains.
"This time, we did a best-practices process--really a worldwide assessment of possible vendors--and wound up choosing Comau Pico [Southfield, MI] as the assembly systems supplier," he continues. "They had the automotive background, and we tried to blend automotive-type assembly systems with our medium-volume manufacturing operation. Our takt times are much longer than automotive assembly takt times, so we had to make sure where we automated or put technology in place that it was appropriate for our volumes. We couldn't put a fully automated line in, and we can't complete an engine every 18 seconds, or we'd over-produce."
Based on an overhead monorail and a system of "J-hooks," the 125,000 ft2 (11,600 m2), continuous-flow line can produce all Verado engine variants in lots of one, with no changeover. Counting left- and right-hand propeller rotation options, various overall lengths, horsepower levels, and Mercury and Mariner models, there are more than 80 possible variants.
Developed with Comau Pico, the assembly system is controlled by seven personal computers on the power head line and two PCs on the J-hook line. Comau Pico and Mercury engineers also worked together to develop the J-hook and pallet system, which allows operators to easily manipulate the work to the correct height and orientation.
Mercury engineers decoupled the power head line from the final assembly line to enable creation of a power head buffer or "supermarket." Final engine assembly begins when a customer order for a specific model is generated. A barcode for the specific product being built is scanned in on a particular J-hook, and that data follows the engine through assembly to cue each station to the correct process. Seven PCs control the head and block line, while two PCs control the J-hook line.
The engine midsection, held on the J-hook, meets the head and block line for power head mating. This is followed by supercharger and electrical assembly, fuel and air systems, and attachment of the gear case. After filling with oil, the engine then is subjected to cold and hot testing. From that point it's a matter of adding cowling, decals, and other final assembly items.
All fasteners used in the engines are installed using DC digital tools. "There are no pneumatic tools on the line for assembly---there are one or two for other purposes," Oswald says. "Using DC tools allows us to capture an electronic signal to provide immediate feedback to operators, and it reduces noise on the line tremendously--with pneumatic tools, no matter how hard you try to fix leaks there's always noise."
Supplied by Atlas Copco Tools and Assembly Systems (Auburn Hills, MI), the DC torque wrenches also allow operators to perform tasks such as tightening to a specified torque, then backing the fastener off 1/4 turn. "It's tough to do that by hand, but with the DC tool it's automatic and we can document that it's been done," Oswald says.
"The DC tooling technology has allowed us to do many things we weren't able to do in the past," he continues. "We can program the torque profile to avoid overshoot by looking for multiple torque 'milestones' along the way to the specified final torque. For example, if we want a final torque of 24 ft-lb [33N*m], we can program the gun to go to maybe 15, slow down, and go toward 24 at a slower rpm. We can graph what that torque profile looks like, and we can see whether it's right or not.
"We've used that capability on a number of occasions for analysis of engine failures," Oswald says. "We can see if the torque profile was bad, or if we had a component failure or some other problem."
If more than one fastener is used at a station--for example, in tightening connecting rod bolts--steps are shown sequentially on the operator station display. Robots are used for critical fastener applications or when very high torques are required. "We have a robot for torquing main bolts, because they're critical and it's a pretty high torque," Oswald says. "Fourteen of the main bolts hold the entire engine together, and we went through a lot of changes from an initial torque spec to the production torque strategy."
Another key piece of technology is a pick-by-light system at some stations. An example is crankshaft installation in the engine block. Cranks are marked with a dot-matrix code that includes data on the sizes of all the diameters on the crank. The operator places the crank in the fixture, where the dot matrix is read by a vision system. A system of lights indicates which of 32 bins contains the matching bearings.
"There's a light beam that has to be broken by the operator selecting the appropriate size bearing," Oswald explains. "If two bearings of the same size are needed, the operator has to break the beam twice. So it's not just a pick-by-light system; it actually has an interaction feature to it."
Once correct bearing sizes are selected, a PC display shows the operator where to install the bearings on the crank. Oswald says the system is crucial to minimizing vibration in the engine. A similar setup is used to allow select fitting of tappets based on camshaft dimensions.
Lean thinking also permeates the Verado assembly process. The line makes extensive use of lean concepts, including level scheduling, 5S, TPM, visual factory cues, and mistake-proofing, to allow one-piece flow of all Verado engine variants.
According to Oswald, level scheduling is the foundation upon which Mercury's other lean efforts are built. "You've got to start with scheduling, you've got to pull product through the facility, and you've got to have a level schedule," he says. "That's the beginning of the lean journey, and that's going to raise issues that you can then attack and resolve." Schedules are manually introduced into the line each morning using a card system and a level scheduling (heijunka) rack.
Visual factory cues include a system of lights that allows everyone on the floor to quickly see the status of each station on the line. "We decided we didn't need green lights, since that would mean everything's OK," Oswald explains. "Yellow means there's a problem but it's not critical. Red means the process is down due to quality or equipment problems, and blue indicates the station needs material."
Quality assurance for the Verado line includes several in-line test stations. The first is used to test power heads after assembly. "One of our strategies was to make sure subassemblies were good before we send them along," Oswald says. "So at the power head test station, a laser comes across the top of the valve guides to make sure they're all seated properly. This is followed by leak testing and vibration signature analysis, which involves exercising all the valves, then capturing the resulting vibration data."
Cold testing and hot testing take place on all engines after most assembly is complete, and both test stations are in-line. For cold testing, the test system is connected directly to the engine flywheel, which is rotated to check the fuel system for leaks, and pressurize cylinders. Cold testing consists of more than 70 different checks, including static and dynamic vibration analysis. Engines that pass cold testing go to hot test; rejects are sent to rework.
The hot testing apparatus connects directly to the engine's electronic control module to cycle the engine through a routine at various engine speeds. In-line hot testing results in significant reductions in cycle time compared with the previous method of queuing engines for test, but Mercury Marine is still not satisfied.
"We have a specific plan to step down from 100% hot testing in 10% increments," Oswald says. "We're monitoring first-pass yield, and if it gets above 99%, we're going to audit mode on hot testing. We're very close to that now."
Oswald says the company would continue to hot test 90% of engines at first, stepping down eventually all the way to 10%. "Once we're in audit mode, we can always revert back to some amount of hot testing--all the way to 100% if we have a problem. That way, we can analyze the problem and get to the root cause," he explains.
"Every step down in hot testing is 10% more capacity," he adds. "If we get down to hot testing one engine in 10, then have a problem that causes us to go back to 100%, you can bet there'll be some intense pressure to get that problem resolved. And that's part of the philosophy: don't hide the issues, don't work around them, get them exposed and address them."
Mercury is also using technology to improve traceability of engines and key components. "We have serial numbers on our other products, and we have some ability to trace lots. But on this product, we're collecting 'birth history' at each station," Oswald explains. "We collect information on the number of torques, and other electronic info such as a required oiling operation, use of pick-by-light systems, and so on. We identify key components and either scan those in or use digital technology to capture that data. So now we have a complete birth history, and at the end we upload the info to a server and save it." The resulting data allow Mercury engineers to analyze the assembly process, and also can be accessed and used by field repair technicians and other groups, he adds.
"There were a lot of targeted goals we wanted to reach with this line, and we are hitting most of them," Oswald says. "We had quality goals, safety goals, and ergonomic goals.
But Mercury's main goal was improved throughput with one-piece flow, and the line is achieving that even as a second shift and production of four-cylinder engine variants ramp up.
"On our older, high-horsepower V-6 products, cycle time could be up to 120 hours from the start of a power head to the box. Now, we can put an end-model-specific order in the system, and we can box the engine two hours later. So we've significantly reduced throughput time," Oswald concludes.
"A Game-Changing Product"
In the outboard market, the battle for dominance is between two- and four-stroke engines. Each has advantages and limitations: two-strokes are powerful, but have relatively high emissions, noise, and vibration. Four-strokes are cleaner and quieter, but have historically lacked the low-end torque needed to quickly get a boat on plane or efficiently tow skiers.
"We were challenged to combine low-end torque with fuel efficiency, reduced vibration, noise, and emissions, and good handling," Mike Oswald recalls. "You can try to clean up the two-strokes, and we did that with our OptiMax product line. They're very low-emission engines even though they're two strokes, and they have very good performance.
"But the thing you get with a four-stroke that you don't with a two-stroke engine is improved noise-vibration-harshness [NVH] characteristics. Typically, in the outboard engine market, you trade off performance and NVH characteristics," he says.
Mercury's solution was Verado--a line of supercharged, intercooled four-stroke engines that Oswald says constitutes "a game-changing product" for the outboard industry. The double-overhead-cam engines are available in seven horsepower levels: 200, 225, 250, and 275-hp in-line six-cylinder models, and 135-, 150-, and 175-hp in-line four-cylinders.
The engines are the only supercharged four-strokes available in the outboard market. They also employ Mercury Marine's power steering and SmartCraft digital throttle and shift, which improves throttle response by eliminating long mechanical cables running from the helm to the engine.
This article was first published in the April 2005 edition of Manufacturing Engineering magazine.