Configuration Catastrophes – V

Introduction: This is a classic design fault that we have all seen many times, but they somehow keep making them - sometimes in new and interesting ways, but usually with the old classics. They say that the best conmen find marks who want to con themselves, and then help them do it. For people who don’t want to pay full price or who want to take shortcuts, “V” is their desire and their punishment. What is “V”?

That isn’t “V”, but someone needed to make this joke, and I volunteered. “V” refers to the shape of the redundancy concept (picture later). A V-shaped torso may be attractive, but ever notice how really strong people are built more like boxes? A V-shaped redundancy concept is attractive, because it is broad at the top and looks like it can handle a lot, but it is inexpensive to buy because it tappers away to almost nothing where the support services are needed to make it work. This means that the top redundancy concept isn’t real or supported. No one forces people to buy “V”, it works more like this:

Emperor’s Clothes: If anything is too good to be true, it usually is, but that doesn’t stop shipyards selling V-shaped designs to vessel owners, who peddle them to vessel clients. Shipyards do it because owners demand it, and owners do it because their clients demand it. What use is an honest 2 or 3 split, when people fool themselves into thinking they can get “genuine” 6, 8, or 10 splits for the same price? Thus, the race to build the most stupid designs is underway. We need the maximum number of redundancy groups for the minimum price and something has to give. The emperor is wearing no clothes, and many multiple split vessels have no redundancy groups.

Zero Groups? They have no redundancy groups, because a redundancy group is self-contained, self-sustaining, and capable of controlling vessel surge, sway, and yaw. Many of these designs’ “redundancy groups” are just grouplets without independent support services or the ability to control vessel position on their own. The theory is that only one of the many grouplets will fail and all the remaining grouplets will maintain position and services. The constrained reality is often different. Why is this attractive?

Of Groups & Grouplets: If you had 6 DGs and 6 thrusters, you can have 2 real redundancy groups, but are left with 3 DGs & 3 thrusters after the worst case failure of one independent group. With 3 real redundancy groups, the worst case failure of one independent group leaves 4DGs and 4 thrusters – a 33% improvement, but at the cost of an extra set of services. For 6 real redundancy groups, you need to double the number of thrusters because you need one forward and one aft, and double or triple the number of support service systems compared to 3 or 2 split. This is very expensive in money and space and only gives a 25% improvement over the 3 split. If only there was some way to get all the benefits of 6 split without paying that hideous price in support services and thrusters. Enter grouplets accompanied by CBT.  

They’re Flying! At the main switchboard level, the grouplet design looks six split, if they can make and keep closed bus tie operation redundant, but when you look below the main switchboard to the support services, there is often nothing to carry the overbearing grouplet split concept and they are often 2 split or less. Even if that wasn’t the problem, real redundant closed bus tie operation is so expensive to implement, and especially to maintain, that if owners and clients knew how much it would cost them in lost time and extra work, they would go to simple splits.

Back to V: The picture above gives a real example of a typical V-shaped design. It’s broad at the top where the customer facing, medium voltage, main switchboards, which supply the thrusters and industrial mission, are 4 split. The low voltage, ship service switchboards, which support all the vital services that the 4 split depends on, are only two split. This could work if the vital pumps were duplicated, but the money was spent on increasing the number of grouplets, so two split is the reality. It could have been worse, they could have fed each vital pump from an automatic changeover switch (ACOS) and made everything common. Speaking of common, the bottom of the V shows the common day tank, which supplies all the diesel generators that supply the top level, so four split is bunk and everything is common. But it’s approved.

Common: This isn’t unusual. Once you start looking for V-shaped redundancy designs, you start finding them everywhere. The weird thing is that most people don’t see them. The top level looks attractive for customers and analysis often stops there. The less interesting support levels often have fewer redundancy groups. Sometime this is OK, such as with properly combined, protected, and maintained control power, but often you end up with designs that don’t have enough settling tanks, day tanks, purifiers, cooling systems, compressed air groups, etc. to possibly support the higher redundancy concept. Sometimes, this is a mistake, and sometimes it is a gamble. When pneumatic controls were common people knew to split compressed air systems or add air reservoirs to DP vital control loads, but now that pneumatic controls are rarer, even class reviewers may not notice the lack of safety and need to be fought to fix it. As an electrical engineer, who used to work as a reliability engineer, I see a lot of bad electrical shortcuts being used to make up for too few redundancy groups, but most of them don’t seem to get caught. The more you look at support systems, the more interconnections you find. Like this design:

Detection: If the concern is the V, then the solution would be to ensure that the high level redundancy concept is supported by each of the lower levels. It should be box-like not V-shaped. I developed a simple tool for this 15 years ago. A simple, color coded, redundancy matrix lists the redundant main equipment across the top, lists the vital support services needed in each row, and fills each cell with color coded information reflecting the services. These tables were originally developed to catch redundancy mismatches, but are also useful for detecting V designs. The smaller, simplified DNV and OCIMF versions are less likely to catch these problems, as the information is scattered across multiple tables rather than revealed in vessel wide tables. The tables are further discussed in another article and a link to the table instructions given there.

V is here and here to stay,

Until caught, won’t go away,

With your expectations plays,

Reality keeps him away.

Conclusion: Don’t be fooled by V. Let the reality of the equipment and systems define the real redundancy concept, rather than bait and switch sales hype and its supporting approved documentation. Use the redundancy matrixes to help you see.