On Dogs and Ferries

I like to take examples from one field and see if they give lessons for other fields, sort of the "case history" approach so popular in MBA programs. One concept I insist upon is optimizing design for environment and task, and there is a bit to this, especially with the increasing capability of Computer Aided Design/Computer Aided Manufacturing (CAD/CAM).

Perhaps the best and certainly the oldest reminder of the advantages of design for environment and mission is the diversity of dog breeds. The Nova Scotia Duck Tolling Retriever, for example, is not only suited for the Canadian Maritime climate, but also is specifically adapted for Nova Scotia hunting practices and waterfowl species: It "dances" along the shoreline to entice (toll) curious ducks into the hunter’s range and is small enough to work from a canoe.

My wife and I chose a "Toller" as a companion animal because they have the genial disposition and the intelligence of Labs and Goldens, and a playful nature due to their "tolling" behavior, but are not so large - they are the smallest of the retrievers. However, we had to accept that the "Nova Scotia" part gives a thick double coat and furred feet, which means lots of grooming. We now have a West Highland White Terrier, a "Westie". The Westie’s white coat was developed for a purely utilitarian purpose – Westies were bred to flush small game, and their color prevented them from being accidentally mistaken for game and getting shot. They have adapted well to their current companion role – they are small, playful, cheerful, and quite biddable, at least by terrier standards. However, again their original design comes into play – "terrier" refers to the earth, and more specifically aggressively digging to raise small game. Terriers are also very scent oriented, and like most scenting dogs, they role in interesting scents, presumably to keep them for later reference. Thus, a Westie’s idea of a great end to a day of digging in the mud and rolling in various rotten, smelly things is to leave little footprints on the rug on their way to leaping into your lap. Dogs aren't available other than "off the shelf" or maybe "proven parent" (literally), so when a dog is outside its original intended environment and mission, there are compromises. You can't get a dog custom designed to your exact wishes.

What does this have to do with ferries (and yachts, and military craft, and workboats …)?

One of my great frustrations as a naval architect is that so many boat owners insist on "off the shelf" designs or "proven parent", (or more often "proven parent, but"). In some cases, it is even required by law, and in the coming months we will probably see this kind of thing for new ferries in San Francisco. This is understandable, but wrong. People are used to buying things like cars or computers by going to the store and selecting one "off the shelf". Such products are made in huge numbers so it is feasible to spend lots of money on machinery and tooling to reduce the cost of making them through mass production techniques like automation – the tooling for a new model SUV is typically over a billion dollars, but the factory builds one every two minutes.

Boats, especially commercial boats, are not made in large numbers, though, and there aren't too many "Ferries 'R Us" outlets. The 2004 WorkBoat magazine's annual survey of new construction listed a total of 442 self-propelled commercial and military boats and small ships built by 45 different shipyards, so each yard only averaged ten boats. Even in any one shipyard, this comprises several different designs, so it isn't common to see more than three boats built to any one design. (One of the few large orders that year was for remote controlled target boats - for some reason they are needed in large numbers.)

A well-proven method of estimating the cost reduction in shipbuilding for follow-on vessels suggests that after about a twenty percent reduction from the first to the second, (which covers engineering, initial setup, bid preparation and so on), the cost drops by four percent each time production doubles, and this only applies to certain categories of costs (metal, for example, doesn't get any cheaper once you have bought about half a boat's worth). Thus for even thirty identical vessels, the average cost for the first two dropped to 89% of the first, but the average for 30 is still 75% of the cost for the first one. Even this doesn't work out to a lot of savings in the overall picture. The first cost of the ship itself is a relatively small part of the lifetime operating expenses (just like a dog, oddly enough), because this cost is spread over the entire life of the vessel. At a recent workshop on ferry economics, it was pointed out that three employees added as much expense as $1.5 million of vessel first cost.

The obvious conclusion from this is that it doesn't take much optimizing to beat any economics of mass production. A relatively small savings in fuel, for example, covers a lot of first cost - saving one gallon per hour (probably less than 2%) is worth nearly $100,000 in first costs. Since the key to optimizing is designing for the specific service and environment, an "off-the-shelf" boat is probably a false economy. For example, a very important aspect of optimization is speed. There are differences between a twenty knot hull and a twenty-five knot hull that make a difference in power and fuel, and a boat designed for twenty-five knots will use more fuel at twenty knots than one designed for twenty knots in the first place.

This is especially the case for commercial craft that have to be meet regulations. Various regulations can make a big difference in operating costs. Each country has some peculiarities that provide hard restrictions on the design. By exceeding a certain parameter a boat may change from one class of regulation to another, resulting in a major change. US rules, for example, have a break at 100 gross tons admeasurement from a "small passenger vessel", essentially a boat, to a large one, essentially a ship. Admeasured tonnage is the internal space of the vessel with certain peculiar exceptions and exemptions (under the unique US system) that can be exploited by a knowledgeable designer. Nest time you are on a ferry, look around for a panel held on by strange looking bolts, probably at the rear of the vessel. It may be marked "tonnage door – keep clear". This door exempts the volume it encloses and as far as the rules are concerned you are sitting in a "temporary covered space", none of which counts for tonnage. If you could get inside the hull, you would find some special very large frames. The space outboard of the inner face of these frames doesn't count either. Even the way the stern of some ferries slants forward is often part of a scheme to reduce admeasured tonnage; the door trick works best if the deck above the door is not the longest. Tonnage dictates the number of crew, their licenses and many other aspects of operation and construction cost, so most ferries are very close to 99.9 gross tons (find the certificate of inspection – it’s posted somewhere in the main passenger area). A design intended for some other country will have been designed to another set of rules and will have to be modified to achieve the savings from these peculiarities (and to eliminate weird things to meet the regulations it was designed to) and this can have other consequences that move the design off an optimum.

Owners also fear that a new design is somehow unproven and high risk. This is only true if the whole concept is entirely new, but like new dog breeds, most saltations in fast craft design are hybrids of one sort or another. They thus can be analyzed pretty reliably from data on the pieces that were combined. A friend and I have developed a new design concept for a high-speed hybrid planing hull that uses aft mounted hydrofoils. (Some people may remember a sort of dumpster with wings undergoing sea trials on the southern San Francisco waterfront a few years ago.) This concept is sufficiently new that I would be reluctant to advise an owner to adopt it without some significant trials and research, but it was still reasonably predictable from combining hydrofoil data and planing craft data. It is also important to realize that there isn't much new under the sun as regards ship design - after all, we naval architects think of either Noah or God as the first ship designer (too bad the discussions about contract change orders have been lost). In the case of our radical design concept, when we sought a patent, we found that it had been patented over fifty years ago, and hybrid hydrofoils may have seen service in the German Navy in World War II.

The design of just about any monohull or catamaran is pretty routine, especially with current techniques of computer aided engineering. I have to reluctantly admit that naval architects are just another flavor of engineer, not some sort of maritime Harry Potter, and that ship design is just routine engineering. (I always wonder if people who want an off-the-shelf boat also want an off-the-shelf building, freeway or bridge.)

Most designers use an integrated design system that allows exact definition of the hull shape, which is then transferred to structural, hydrostatics and hydrodynamic analysis software to verify safety and performance. Modern structural software can then automatically optimize hull structure for weight or cost (which in turn depends on the shipyard's practices) as well as ensuring freedom from failure. Finally, the ultimate (and usual) proof is model tests. A scale model is run in an instrumented basin. These tests can be done with or without waves, and can model steering and waves and wind. It is expensive - typically a good test program might cost $50,000 or more, but that is a fraction of the cost of the boat, and less than the cost of that gallon of fuel per hour. It is worth noting that recent tests I ordered for a modification to an existing patrol craft produce enough fuel savings to pay off the tests and the modification in less than three years.

Actually, a proven design is often higher risk than a new one, because it is usually "proven design, but". An owner likes a boat already in service with some other owner and wants one just like it, with just a "few minor changes". The changes usually add weight, adversely impacting speed and stability, or they can have other unforeseen consequences, but the design is "proven", so no one checks it out carefully. One of my favorites was an owner who needed a large hydraulic winch (among other things) added to the aft deck of a workboat. The parent design didn't have enough room to put the new power pack in the equipment space aft of the living quarters, so it had to go in the engine room forward of the living quarters. The high-pressure hydraulic lines ran through the overhead of the captain’s cabin, and of course, they constantly leaked. Weight additions especially are a very high risk for higher speed craft, which depend on dynamic lift – fast craft are weight sensitive, and it is easy to add just a little bit too much weight so the overloaded hull loses speed all out of proportion to the added weight.

The computer has also reduced the cost of specialized design and construction. The same model developed for analysis is transferred to a detailed structural and equipment design package, which provides all sort of automatic tools for doing the routine development of the structure, piping and other details. The result is a three-dimensional "product model". Before modern drafting was invented, an "Admiralty Model", an actual scale model, was used to develop the wooden ships for that period's "Iron Men". We have gone full circle and now "Silicon Person" has replaced paper drawings with an electronic 3D scale Admiralty model. However, the new model "talks" directly to computer controlled equipment that cuts metal (or foam for fiberglass molds) automatically. The model also provides data to material and work control software and does all the various e-business stuff (ship design and construction collaboration over the Internet is now routine). There are lots of other improvements, mostly enabled by computers, that have radically improved productivity as well, and any reader not yet bored can find out more in the Journal of Ship Production, Transactions of the Society of Naval Architects and Marine Engineers or Marine Technology. However, this all means that the cost of a boat to a new design is much lower than it used to be. One shipyard, which was already using computer steel cutting, reduced construction labor costs by 20% by adopting a product model and other improvements on the first 160' offshore supply vessel (OSV) they used it for - without any increase in design costs.

The bottom line is that if an existing vessel or design provides exactly what you want, fine, but otherwise, don't compromise and don't be afraid to get just what you want. It won't be very expensive or risky in the short run and will be well worth it in the long run.

This also has much larger implications for business, especially as regards the impact of CAD/CAM. We have the potential of a new age of customization, due to computers, and flexible automation. It is now feasible to manufacture certain types of objects as largely custom designs, or at least as highly customized ones. This latter case is worth exploring just a bit, again in a marine context. I was chief engineer of Munson Manufacturing, a shipyard building aluminum workboats. The method Munson used was to design a custom boat within the constraints of a parameterized hull form that was readily constructable on the yard’s flexible tooling system. The hull parts were designed in 3D CAD and computer cut, and all of the systems were standardized on a basic level of components, details and interconnections. (Everything was also parameterized in a computerized bid and ordering system.) This enabled a custom boat, that suit the owners needs very well, to be built at a low price by simply assembling standard parts.

I can see this as being readily applicable to a wide range of products where high levels of customization provide high enough increased quality (where quality is defined as best meeting the customer’s needs) to be worth the additional cost. However, clever use of CAD/CAM and related technology would also reduce the cost. Two examples that are immediately feasible are clothing and medical devices, (though I recently saw a challenge on the Food Network where one competitor used a CNC milling machine to make a gingerbread house).

In the case of clothing, there are any number of sources for reasonably affordable scanning devices which can produce a three dimensional model of an individual. (I have used a Faro Arm capable of digitizing a person for measuring propellers – it cost about $15,000, but its accuracy of 0.0001" is probably excessive for measuring people, and even then it’s probably affordable.) Once a person was digitized, a garment designed on a standard size could be parametrically morphed so it fit the individual, and it could be adjusted as required to make it look and feel right, lapel width proportionately adjusted and so forth. It could be tried on virtually, and cloth and color selected, and then a computer driven cutter would cut all the parts. It probably would cost more to do the sewing locally, but this might well even be covered by reduced shipping and handling costs, and especially layers of production, wholesaling, advertising overhead and so on. In fact, since a custom garment would be guaranteed a sale at the full price, the cost of all those discounted items in end-of-season sales (and worse for profits, showing up at TJ Max) would be eliminated. We could see a return of the "bespoke" suit, and this would certainly be a boon for some women who could get clothes that fit well and looked the way they want, without worrying about what is on the shelves just because teenage girls think it’s cool.

A more important application would be medical devices. Items such as hip and knee replacements would be custom made from a parameterized model (available in design packages like AutoCAD Inventor and Solidworks), developed by digitizing data from CAT scans and physical measurements, instead of being chosen from a series of premade, sized products. A three dimensional printer (a sort of ink jet device that spits out plastic instead, and does it in three dimensions) would make plastic parts that could be used to make molds, or molds directly. In the case of metal parts, most stainless steel medical devices are investment cast anyway, so the 3D printer would make the patterns. The data for the part would go over the Internet remotely, and the part would be sent out a day or so later. (We got our precut metal from the cutting service 200 miles away two days after we uploaded the computer data to their site.)

In the long run, clever design approaches, especially using the concept of the 3D product model tightly linked to analysis tools, could revolutionize a wide range of products. This would not only make the product better, but it could bring back manufacturing jobs – This would also counteract what I see as a depressing tendency towards homogenization in products, let's call it the Wal-Mart effect: The best products might be more expensive than those mass produced overseas, but their better fit to the specific customer’s needs would justify the cost, and in the long run, the ultimate prestige item would not be a brand, but a custom, one of a kind.

Readers interested in more information on modern ship production can go to http://www.sname.org or http://www.NSRP.org. A typical suite of ship design software can be seen at http://www.shipconstructor.com. Those interested in Nova Scotia Duck Tolling Retrievers can go to http://www.nsdtr-usa.com or http://toller1.com .