Imagine yourself safely walled inside your 12th century castle. Its walls are sturdy enough to survive a pounding from an enemy using a traction trebuchet — a technology that had been around for more than 1,500 years — and could hurl a 250-pound stone up to 200 feet. Imagine peering from those walls at a new technology, the counterweight trebuchet, capable of hurling larger projectiles a far greater distance or that same stone more than four times as far. By skill or luck you manage to survive the onslaught and recover one of these new weapons. In addition to repairing walls and tending to the injured, your top priority is reverse engineering this new weapon-for defensive or offensive purposes.
Fast forward towards the end of WWII. The Soviet Union was an ally of the US, but there was an element of distrust between the two nations. Although the Soviets desperately wanted a strategic bomber, the US refused to provide the B-29s that Stalin wanted. Then in 1944, four US B29s made emergency landings on Soviet soil. The US demanded their return, but the Soviets refused and flew three repairable planes to Moscow and delivered them to experimental aerospace manufacturer Tupolev OKB. By 1947, the Soviets had completely reverse engineered the B-29s and manufactured the nearly identical Tupolev Tu-4. Nearly 850 Tu-4s were built and flown from the late 1940s to the mid 1960s.
Playing catch-up with military technology may very well be a matter of survival. Whether catapults, crossbows, or strategic bombers, an enemy’s superior weaponry is always desirable, and reverse engineering may be the quickest way to get it. Reverse engineering may be used not only to replicate them, but also to determine a weapon’s Achilles’ heel.
The entity that performs reverse engineering on an existing weapon, product, or system has certain advantages. One of the major advantages is the elimination of prototyping and trial-and-error that may have been part of the original item’s development. The time in man-hours for the Soviets to reverse engineer and manufacturer the Tu-4 was certainly less than for the original development of the B-29.
Those involved in another less scrupulous form of reverse engineering, companies that target high-cost goods such as luxury watches or handbags, also benefit from this advantage. Counterfeiting luxury goods is a very common practice and has given reverse engineering a bad name. And although both illegal and unethical, counterfeit goods are popular among many consumers, who justify purchasing these goods with the belief that the original manufacturer is overcharging. Why should a leather purse cost $1,000 or a stainless steel watch cost $7,000? The consumer ignores product development, warranty, advertising, event sponsorships, and other costs built into the sale price. The party that reverse engineers the goods avoids all of these costs.
Reverse engineering is not confined to small operators. The world’s top corporations reverse engineer their competitors’ products all the time. As you read this, the makers of products from cell phones to automobiles are dismantling the products of their competitors. They’ll probably even acknowledge this is going on.
Reverse engineering a competitor’s product may simply be to draw comparisons that can be used in advertisements. (Although it sure seems that when a new feature is unveiled on a product, that same or similar feature mysteriously appears on the products of competitors). And the number of patent violation lawsuits between the major cell phone makers is mind-boggling.
Reverse engineering is also a hot topic for the nuclear power generation industry and both its negative and positive uses affect nuclear facilities. The negative uses of reverse engineering mainly relate to counterfeit parts such as moulded case circuit breakers; a great deal of attention is being paid to preventing these fraudulent parts from being used.
Obsolescence
The law of supply and demand has had a great effect on the nuclear industry. For several decades the demand for certain parts and equipment has been minimal or nonexistent. When the demand vanished, so did the supply. This has affected different suppliers in different ways. A supplier whose products are used mainly in commercial markets might have simply dropped its nuclear QA programme and focussed on other industries. Suppliers that focussed largely on nuclear struggled. Either they survived to become that last-best buggy-whip manufacturer; or were acquired by a competitor and any redundant products were rendered obsolete; or they adapted and thrived, perhaps by acquiring other suppliers. This latter case is good and bad news for nuclear plants. The good news is that the company that supplied a part 40 years ago is still in business, but the bad news is they rendered that part obsolete 20 years ago.
Manufacturers often are blamed for product obsolescence, but this blame is misguided. Obsolescence is primarily consumer-driven. Why would the manufacturer of a product in high demand stop selling it? The reason why the maker of the SmartPhone 4 launches SmartPhone 5 is to attract the customers it lost when a competitor launched a shinier alternative to the SmartPhone 4. The maker of the SmartPhone 5 doesn’t necessarily want to make the SmartPhone 4 obsolete; they want to make the competitor’s phone obsolete!
One use of the term ‘obsolete’ describes items a plant would like to purchase, but cannot for one of the following reasons: 1) The original supplier no longer exists and the intellectual property was lost; 2) The owner of the intellectual property simply refuses or cannot manufacture the items because materials or sub-components are no longer available; or 3) The owner of the intellectual property quotes a price and/or lead-time unacceptable to the purchaser.
Some people may challenge the validity of the third above reason for obsolescence. But by that same token, we human beings still have the capacity to manufacture telegraphs, vacuum tubes, and wind-up phonographs. Nothing is truly obsolete, but rather progressively more economically impractical to manufacture. If a nuclear plant in an outage is quoted a 52-week lead-time for something required to restart the plant, for all intents and purposes it is obsolete.
As obsolescence can be largely caused by the lack of consumption, reverse engineering is also driven primarily by the consumer. As with so many other things in this world, the decision to reverse engineer something comes down to money. For example, if the toner cartridges for your home office laser printer became obsolete, one option is to simply purchase a new printer and throw away the old one. The cost of a new printer (which comes with toner) is often comparable to the cost of replacement toner cartridges.
But what about a power supply used for control room instrumentation in a nuclear power plant? Assuming that a new power supply was available, a great deal of engineering would be required to install it. Plus the new unit would need to be qualified, and most likely modifications would need to be made to connections. The cost of the engineering and modifications would likely be many times the cost of the power supply.
Reverse engineering a power supply is also expensive, and the resulting new unit also would need to be qualified. And given that the new unit will likely be different, it too will require engineering to accept. Why then are so many obsolete power supplies reverse engineered? There are two main reasons and both relate to money.
The first reason is that power supplies available today look nothing like those manufactured a few decades ago. New units likely contain some digital technology and are physically smaller than the old units. Therefore, the engineering required is likely a design change that can easily cost a few hundred thousand dollars. The second reason applies to sites that have the same obsolete power supply installed in multiple applications. Nuclear sites avoid having two different types of equipment performing the same or similar function, as this may be confusing to operators. Therefore, the decision to use a new power supply means replacing multiple units, which of course, is many times as expensive.
Although reverse engineering a power supply and manufacturing a replacement is a costly and time-consuming endeavour, replacement units can be made to appear identical, interface to the plant using the same connections, and function in a similar enough manner as to be considered equivalent replacements. They will still require engineering, but an equivalency evaluation is far less costly than a design change, and the plant does not need to replace multiple power supplies.
The same logic applies to other kinds of equipment. For example, a plant may have a large pump that is obsolete, but needs several replacement parts to perform routine maintenance. These parts may be simple static parts such as fasteners, or more complex dynamic parts such as impellers. Both these types of spare parts are routinely reverse engineered for both nuclear and commercial applications. The reverse engineered parts may be more costly than original parts supplied decades ago, but the cost of reverse engineering plus an equivalency evaluation is far less expensive than a complete new pump plus a design change.
The engineering process
In order to understand the reverse engineering process, the (normal) engineering process needs to be explained. The process begins with a stated need; a better smart-phone, a more powerful engine with increased efficiency, a breaker that requires less maintenance. The next step is determining the parameters. What is the expected performance? What are the constraints (size, cost, weight, materials of construction)? What is the operating environment? With the expectations defined, the next step is to develop possible solutions. These possible solutions need to be evaluated for manufacturability and cost. A concept for a more efficient engine using titanium for the major components will not be a practical solution for a production automobile. Practical solutions typically lead to the manufacture of prototypes, which are tested, evaluated, re-engineered, improved, and so on until a final specification is developed. This final specification is then used for production.
The very fact that reverse engineering is being considered means you do not have the final specification. If you had the final specification, you could simply manufacture new parts. If certain sub-components were unavailable, you could re-engineer the design, but that’s different from reverse engineering (see below). If you are considering reverse engineering, you likely have a sample of the part or component. Without a sample, there is little that can be done. Preferably the sample is new or in good working condition. Any documentation you have will help in the process. Operating instructions or maintenance procedures help immensely.
While the primary reason we perform reverse engineering in the nuclear industry is to obtain obsolete parts and components, the end product of reverse engineering is exactly the same as the end process of engineering – a specification that can be used to manufacture new parts or components.
The goal of reverse engineering is to produce parts or components that are identical to the original. This is a little more practical with mechanical parts, though 100 percent compliance to the original specification is virtually impossible for reasons discussed in greater detail later.
At first glance, mechanical parts appear to be very simple to reverse engineer. It’s just like duplicating keys, right? But reverse engineering mechanical parts requires much more than just replicating the shape. Mechanical parts are most often used in conjunction with other parts. If a part is threaded, what does it thread into (or onto)? Is the part used in a static or dynamic application? Will it be subject to fatigue? Is it part of a balanced set? The manufacturing specification for the original part will have tolerances on all dimensions. Lacking knowledge of those tolerances, there are several ways to estimate them.
■ Variations between samples, if multiple samples can be obtained. If a feature on one original part measures 5.375 and another measures 5.312, you could establish the new range at 5.312 to 5.375, which is huge using modern manufacturing techniques.
■ Variations within a sample. If a feature measures 5.372 in one area and 5.369 in another area of the same feature, you could use those measurements as the new range. This is almost certainly tighter than the original requirements, but still possible to obtain using modern manufacturing techniques.
■ Engineering judgment. Given the numbers 5.372 and 5.369 from the previous example, one might conclude that the target maximum was a standard fraction such as 5 3/8 inches or 5.375. You could then make the new range 5.369 to 5.375, which is double that of the previous example. And depending on the type of feature, perhaps 5.375 was the target with a plus or minus tolerance.
■ Standard features. A non-standard part may incorporate standard features. Threads are a good example of this. A design engineer would be foolish to invent his own thread design when perfectly good designs are available. If certain features conform to a standard, that standard was most likely referenced in the original specification.
■ Mating parts. When parts interface with other parts, the uncertainty is increased. If one part fits into a second part, the maximum dimension of the first part must be less than the minimum dimension of the second part. The difference is called the allowance. When attempting to determine one of these dimensions, it certainly helps to know the other; having both parts in your possession helps determine whether this is a loose or tight fit. If the fit is extremely close, it would be advisable to obtain additional samples. Another option would be to reverse engineer both parts as a set.
It is fairly important that the organization that performs the reverse engineering is knowledgeable in the manufacture of similar parts and components. A manufacturer may recognize a special process, such as forging, or a special surface finish that may escape the eyes of someone unfamiliar with manufacturing.
Selecting material specifications is another area where manufacturing expertise is valuable. Most material specifications have fairly wide ranges for chemical composition and any given sample may comply with multiple specifications. Mechanical properties have even wider ranges and are most often expressed as minimum tensile strengths with maximum hardness values. Extrapolating hardness to tensile will result in huge ranges; typically manufacturers aim for a middle point between the limits. This means whatever tensile strength is determined from testing the sample, the original specification likely required a minimum easily 10 to 20 percent lower. Therefore, new parts manufactured to the specification you choose could be 20 to 40 percent stronger and harder than the original. Using what is likely the mid-range strength as the new required minimum may seem like a conservative choice, but it is not.
Electrical and electronic parts
Although determining tolerances and material specification can be a big challenge with mechanical parts, manufacturing the replacement parts is relatively simple. In contrast, electrical and electronic parts are typically more complex and incorporate discrete components that may themselves be obsolete. It is one thing to reverse engineer a control board, but entirely another to reverse engineer and manufacture a diode or silicon-controlled rectifier.
Given the likelihood of one of more discrete components being unavailable, the reverse engineering process for electrical and electronic components is typically followed by re-engineering (or simply engineering). The reverse engineering process determines what the existing component is and the re-engineering process results in the design of an equivalent component constructed using available sub-components.
Re-engineering brings equipment qualification into the picture. If the new component design contains different parts and materials, the supplier likely needs to perform EMI/RFI (electromagnetic interference/radio frequency interference) testing, thermal aging, and seismic testing.
Under siege
It probably wasn’t very hard for the procurement engineering and supply chain professionals at nuclear plants to imagine themselves in a 12th century castle being surrounded by the enemy and seeing the defences pounded by stone-hurling trebuchets. You’re offered few choices and none of them are ideal. Reverse engineering is not ideal either. But given the right set of circumstances, reverse engineering is a powerful tool in the never-ending battle to keep plants operating reliably and in as close to the original configuration as possible.
Legal Aspects |
| The vast majority of the items reverse engineered for nuclear plants are not covered by patents that apply to the specific items. One reason is the age of the items, which typically exceeds the length that patent protection applies (often 20 years). The second reason is that most reverse engineered items are not patented to begin with. The third reason is that in the USA at least, patent infringements are battled in civil, rather than criminal, court. It is up to the patent holder to claim that a patent has been infringed upon. Or, while it is unlikely that a thirty-year-old impeller or power supply contains anything protected by patent, the original manufacturer may claim their specific design is proprietary. It makes almost no sense for a company to spend money on legal fees to bring a patent infringement claim against another party over a product that has little economic value to the patent holder. It would be very difficult for such a claimant to argue they have suffered damages in the form of lost revenue on a product they no longer sell. |
Bailey 862 circuit boards |
| NLI was asked to provide replacement modules for approximately 900 obsolete Bailey 862 modules for a nuclear site. In addition to replicating all of the original functionality of the modules, the site requested several performance enhancements. Sites often refurbish boards such as this, but there is a finite number of times refurbishment can be performed. In addition, if discrete components need to be replaced, the availability of those components is obviously a factor. After reverse engineering was complete on the original modules, NLI began re-engineering the replacements. It should come as no surprise that many of the replacement components were smaller than the original. This allowed NLI to put all of the individual components on one single board versus the stacked two-board configuration in the original module. |
Greg Keller, strategic business development manager, AZZ | NLI, 7410 Pebble Drive, Fort Worth, Texas 76118
