120 tonne GRP tanks shop fabricated

Plasticos Industriales de Tampico (PITSA), of Tampico, Mexico, was awarded an exceptionally challenging project in late 2007. PITSA accepted orders requiring the company to fabricate, within a few months time, four large diameter glass reinforced plastic (GRP) tanks at its factory and deliver them to the port to be transported by heavy lift vessel to a nickel extraction plant in the Pacific Rim. The tanks were 10 m inside diameter (ID) x 15 m, 14 m ID x 8 m and two of 14 m ID x 18 m.

All four tanks were domed top, flat bottom and designed for liquid fill, internal pressure of 5 kPa, vacuum of 3 kPa, wind and seismic. Vacuum of 3 kPa required that the flat bottoms be cored and the shells be stiffened. Vacuum also required the large domed tops to be significantly thicker than normal.

The 14 m x 18 m tanks were fabricated from Ashland’s Derakane 470 resin and Owens Corning’s Advantex E-CR glass reinforcement. The tanks were weighed by the cranes and were surprisingly close in weight due to the careful control of thicknesses. Each tank weighed about 120 tonnes.

GRP – the most practical material

The two big tanks will be used for both hydrochloric acid (HCl) and nickel chloride (NiCl₂) service. GRP was the most cost effective material for these service conditions. While some of the high-end alloys can handle this combined service, the cost would have been significantly higher than GRP. In addition, GRP is relatively friendly for repairs and modifications, given the remote location of the jobsite. Welding exotic alloys has proven to be problematic in remote field locations due to site conditions and the lack of highly qualified welders. GRP laminations can be done using small hand tools and simple enclosures.

GRP is also versatile. GRP can be moulded into complex shapes, which is very helpful when the field operator encounters the need for a special fitting or pipe clip. Clips and brackets can be bonded to a tank without the localised heating created by welding. Metal tanks with coatings and linings typically cannot be welded without damaging the interior. In the case of GRP, clips and brackets can be laminated on, even with the tank filled.

Design, fabrication and testing

The tanks were designed by Fiberglass Engineering Mechanics (FEMech) per American Society of Mechanical Engineers (ASME) standard RTP-1 2005 (Reinforced Thermoset Plastic Corrosion Resistant Equipment) using a combination of Subpart 3A and Subpart 3B methods. The bottom knuckles and the sandwich flat bottoms were designed with non-linear axi-symmetric finite element analysis (FEA) using COSMOSM 2.90 FEA software. Material properties were computed using micromechanics via the well-known Trilam II lamination analysis software.

GRP is a high strength, low modulus material. Modulus can be thought of as the ‘stretchiness’ or ‘spring rate’ of the material. The lower the modulus, the more a material stretches at a given level of stress. Vacuum for large diameter GRP tanks is challenging because of the combination of size and low material modulus.

The cylinders of the 14 m x 18 m tanks each had to be stiffened with three filament wound trapezoidal ribs with a core size of 305 mm high x 178 mm crown x 508 mm base and a 15 mm overwind. Each of these stiffeners weighed in excess of 4 tonnes.

The domed top heads were 2:1 elliptical in order to match existing piping. Normally, 2:1 elliptical heads are only required for significant internal pressure. In this case, the domed head thickness was governed by the 3 kPa vacuum design requirement.

While 3 kPa is not much vacuum for a soda straw, it is a lot of vacuum for a 14 m GRP flat bottom. Significant vacuum in a large diameter tank governs the knuckle design and the flat bottom panel design.

The knuckle is the transition from the cylindrical shell to the flat bottom. In most cases, GRP flat bottom knuckles can be categorised as either so-called hard knuckles, or flex knuckles. Hard knuckles are relatively thick and will tolerate bending loads, while flex knuckles are relatively thin and designed to bend as the shell displaces. In most cases, flex knuckles are not suitable for large diameter GRP tanks subject to vacuum loading. The large forces generated by vacuum can fold up a flex knuckle and peel it out of the shell. For these tanks a hard knuckle design was chosen.

Hard knuckles can be joined to the cylindrical shell either by means of a secondary bond or by winding the shell onto the knuckle. In the case of these tanks, the second bond thickness would be on the order of 115 mm and the bond taper (per RTP-1) approximately 700 mm wide. The bond would be expensive in terms of material and labour costs so the decision was made to wind on the knuckles.

An added benefit to winding on the knuckles is that a shear collar-anchor dog anchorage system can be used as opposed to anchor lugs. A wound-on shear collar can be fabricated in a relatively short period of time, can accommodate two or three times as many dogs as comparative lugs and is versatile in terms of adjustment for bolt position. The downside is that anchor dogs are force-multiplying devices.

Sandwich structures have long been used as a way to support bending loads while minimising material and construction costs. A steel I-beam is a sandwich of sorts. The flanges are the skins and the web is the core.
The British Mosquito bomber developed during World War II used sandwich construction and was a very successful aircraft. In this case, the sandwich panels were made with structural wood skins and balsa wood cores.

GRP sandwich panels are used in a wide range of applications from aircraft to boats to tanks. A GRP sandwich panel design was the only feasible option for the 14 m flat bottoms for this project. PITSA chose to use a polyurethane foam core (from a local supplier). FEMech specified a core density of 160 kg/m3. The GRP skin thickness was on the order of 20 mm thick.

Fortuitous factory location

PITSA is fortunate to be located in the old port area with a large property on a channel which connects to the port itself. When PITSA landed this project, it was necessary to convert the water front property into a large diameter tank and component facility. PITSA cleared the property, poured a concrete pad on the entire area and made a barge dock. It also made a winder building with a trussed removable roof.

As a side note, after the construction was complete, PITSA discovered it should have pigmented the concrete. The concreted area is so large, workers had problems with sunburns due to reflection. Even the famed sombreros of Mexico don’t protect against light reflected from the concrete!

The vessels were so heavy that, once completed, two cranes were required to lift them onto the barge. This requires the cranes to ‘waltz’ the tanks from the assembly pad onto the barge and there is no room for either crane operator to make a mistake.

Once a tank was safely on the barge, it was temporarily anchored down for the short trip to the Tampico port. The tanks were staged at the port waiting for the heavy lift vessel and the subsequent trip to the Pacific Rim.

Since the water front property is located at the main factory, these large diameter tanks were shop fabricated, not field fabricated. PITSA is currently the only company in the world which can shop fabricate and transport huge GRP tanks. Other fabricators have field fabrication facilities located on waterways. But field fabrication is never as efficient and well-controlled as shop fabrication. With its management, clerical, quality control and maintenance departments at the same location as the large diameter facility, PITSA has a marked advantage in terms of efficiency, schedule control and quality control.

Schedule challenges

PITSA faced huge challenges in successfully completing this project. One of the major elements in the success of this project was the PITSA team spirit. Everyone at PITSA was enthusiastic, worked hard and worked long hours. In addition, PITSA brought in outside companies to supplement their own internal resources. For example, there are firms in the Tampico area who fabricate off shore oil platforms. PITSA tapped into these resources for heavy lift expertise and world class steel skid fabrication.

The most difficult problem PITSA faced was the short fabrication schedule. Due to problems with a previous supplier, the nickel extraction plant found itself in a severe time crunch and required PITSA to produce these tanks in a very short period of time. PITSA was able to produce these tanks in about 4.5 months. That works out to an average production rate of about 14 tonnes of laminate per week.

The tough production schedule was made even more arduous by the incredible rainfall Tampico experienced during the period of production. Tampico typically gets about 1 m of rain per year. During much of the production period Tampico was hammered with heavy rains. For example, during one four day period, an incredible 1 m of rain fell.

The most challenging aspects of production were related to the flat bottom. There were three aspects to these challenges: bottom flatness, flipping the knuckle for winding and then flipping the can and knuckle back and bonding the polyurethane core.

Since the tanks were shop fabricated, it was important that the bottoms be very flat so bottom-to-foundation gaps could be minimised. FEMech required a flatness of 6 mm in a 1 m wide annular area around the knuckle and 10 mm in the region interior to the 1 m wide annular area. PITSA employed a heavy construction firm to pour very flat concrete pads. The specified tolerance over 14 m was 3 mm. Working off these very flat pads and paying careful attention to resin shrinkage issues allowed PITSA to come close to the specification flatness. However, the tanks had to be put on cribbing and some filling and grinding done in order to bring the bottoms within spec.

The difficult part of winding on large diameter flat knuckles is in flipping the knuckle over to place it on the winder and then flipping the completed can and knuckle back over. The tank is wound in sections which are often called cans. The bottom can is approximately 50 mm thick and weighs 20 tonnes. It is no small challenge to flip a 14 m diameter part weighing well over 20 tonnes 180°. PITSA paid careful attention to detail and used the manway cutout areas as hinge points to avoid having to cut unnecessary holes in the can or the knuckle.

The most common mistake in tank bottom sandwich fabrication is a failure to get the core bonded to the bottom skin. Typically the bottom skin is first fabricated. Then the core is bonded to the bottom skin in sheets or blocks. After the core is in place, the top skin is laid up directly onto the core. Hence, top skin-to-core bonding is normally easy while bottom skin-to-core bonding is difficult. With a core thickness of about 165 mm, PITSA had to use a well-planned and controlled fabrication process to ensure good bonding.

Computerised winder

PITSA used a mechanical winder for the first two tanks because of schedule but was able to use a FEMech computerised large diameter vertical winder for the 14 m x 18 m tanks. The improvement in quality and productivity was remarkable.

For example, the vacuum ribs on the 14 m x 8 m tank had to be laid up by hand and required 2000 man hours each. Using the FEMech winder, the ribs were filament wound in a mere 3 hours. After having sunk a combined 10 000 man hours into the ribs on the 10 m x 15 m and the 14 m x 8 m tanks, PITSA’s fabrication crew was virtually mesmerised to watch a rib wound in only 3 hours.

A large diameter vertical winder is one of the few machines in the world which can fabricate a product which has many times the value of the winder itself.

Inspection

FEMech was contracted to provide continuous on-site inspection at the factory in Tampico. In addition to monitoring the quality of production, FEMech inspectors provided tips and ideas which were helpful to PITSA.

As one small example, David Davis, an FEMech inspector, showed PITSA a better method for tying winding glass roving. This method significantly reduced the occurrences of roving tangling in the bath. Roving tangling is typically the biggest problem in GRP tank winding.