Steel–Concrete Composite Pressure Vessels for Hydrogen Storage at High Pressures

The proliferation of hydrogen fueled automobiles and other transportation systems in the U.S., and especially in California, as well as worldwide necessitates the need to store large quantities of hydrogen at fueling stations and other central fueling terminals, Fig. 1. One economical way for storing large quantities of hydrogen under high pressure is in a large diameter pressure vessel consisting of a steel inner shell that carries 50% of the applied pressure coupled with a concrete outer shell that caries the remaining 50%. The advantage of such a system is the elimination of requiring very thick cylindrical shells that are hard, if not impossible, to produce at the steel mill and are very expensive to fabricate. Additional cost reduction can be achieved by building the inner steel shell out of layered rather than monoblock construction.

The U.S. Department of Energy (DOE) in conjunction with Oak Ridge National Laboratory (ORNL) undertook a research program with the assistance of Global Engineering & Technology with the following objectives:

1. Obtain a conceptional design for dimensional proportions of steel–concrete vessels at high pressures with optimum cost.

2. Establish a design procedure for steel–concrete composite vessels.

3. Establish design details for venting the hydrogen to the atmosphere in order to prevent damaging the structural steel and concrete components.

4. Design and fabricate a mock up steel–concrete composite vessel with dimensions close to the smaller of the intended field-constructed vessels.

5. Hydrotest the mock-up vessel with water and then cycle the vessel with hydrogen to prove the integrity of the vessel during simulated operation.

A design pressure of 6250 psi (43.1 MPa) was selected to match existing technology for hydrogen dispensing equipment in the field. Various L/D ratios of the shell were investigated for cost comparison. Some of the parameters considered were

1. cost of monoblock versus layered steel shell materials;

2. cost of monoblock manufacturing versus layered shell manufacturing including welding;

3. cost of fabrication of concrete shell including reinforcement and prestressing;

4. ease of access to inside of vessel for installation of instruments;

5. cost of types of supports and lifting configurations;

6. cost of transportations of raw materials and completed vessel;

7. ease of installation of completed vessel in the field;

8. ease of monitoring hydrotesting and hydrogen cycling;

9. design details for venting the hydrogen to the atmosphere without damaging the carbon steel and concrete pressure components.

The cost analysis established the following parameters:

1. An L/D ratio of the shell between 2 and 4 results in an optimum cost.

2. A minimum diameter of 40 in. (1000 mm) is required for manufacturing purposes as well as instrumentation placement on the inside surface of the vessel.

3. A stainless steel liner is needed as a hydrogen barrier.

4. Vessel supports and lifting lugs are to be attached to the steel part of the vessel.

5. The inside steel shell and heads are to be constructed to the rules of the ASME Section VIII, Division 2, pressure vessel code “rules for construction of pressure vessels-alternative rules” as much as practicable. The outside prestressed and reinforced concrete shell is to be constructed to the rules of ACI 318 “building code requirements for structural concrete” and ACI 372R “design and construction of circular wire-and strand-wrapped prestressed concrete structures” as much as practicable.

The design of the steel–concrete vessel consists of the following steps:

1. establish the thickness of the inner steel vessel;

2. adjust the thickness of the outer concrete shell as well as the prestress level at the outside of the concrete shell to maintain the stress levels in the steel and concrete within their allowable stress levels at the design and hydrotest conditions;

3. check discontinuity stresses between the heads and the composite steel–concrete shell;

4. perform a finite element analysis for the total composite construction to check various stress levels;

5. check other miscellaneous items such as thermal, lifting, and transportation stresses.

The steel shell is constructed as a layered shell in accordance with paragraph 4.13 of VIII-2 “design rules for layered vessels.” The thickness of the shell obtained from Eq. (2) is about one-half that required to resist the hoop stress due to internal pressure. The needed additional thickness is furnished by the concrete shell.

Welding of the longitudinal seams of any individual layer in a layered shell causes shrinkage of the welds. This shrinkage results in locked-in residual tensile stress in the individual layer being welded and compressive stress in the layers underneath it. These secondary stresses are normally ignored in commercially fabricated layered vessels since they do not contribute to the stress calculations for determining the required thickness of the layered shell. However, they are important in the steel–concrete composite vessel since they must be combined with the stresses obtained from the prestressing wires on the outside surface of the concrete as well as the stress due to internal pressure in order to obtain the full stress pattern needed for design.

The stress distribution in the steel layered shell due to weld shrinkage of the longitudinal welds of the layers can be formulated from deflection compatibility equations and the stress results obtained from Eqs. (3) and (4) for the inner layer. The stress pattern through the wall of a layered cylinder due to shrink fitting the layers (autofrettaging) is based on the details shown in Fig. 2 as well as the derivations and experimental verifications given by Jawad [1].

Define t_o as the thickness of the outer layer being shrunk and t_i as the thickness of all layers underneath it. Also, define R_o and R_f as the outside and inside radii of the layer being shrunk and R_f and R_i as the outside and inside radii of all layers underneath it as shown in Fig. 2. Define w as the width of the weld seam in the outer layer and n as the number of weld seams in the outer layer. Define p_f as the interface pressure between the outer layer and all layers underneath it.

The value of the weld shrink factor K in Eq. (7) depends on many variables such as weld width and thickness, weld process, and number of weld passes. The value of K is obtained by trial and error as explained below.

The above derivations are based on the following assumptions:

1. The width w of the weld seams is the same in all layers.

2. Weld parameters such as voltage, amperes, and speed are constant from weld to weld such that the shrinkage constant K does not vary from weld to weld or layer to layer.

3. The number of welds n in each layer is constant throughout the layers.

4. The modulus of elasticity, E_s, is the same for all layers.

The actual interface pressure p_f in Eq. (12) is based on a selected value of constant K. Once the pressure p_f is known, then the circumferential stresses in the outer and inner cylinders are obtained from the following equations:

The value of K in Eq. (12) is determined as follows:

1. The values of circumferential compressive stress in the inner layer due to wrapping each of the outer layers as obtained from the strain gage readings and Eqs. (3) and (4) are plotted as shown in Fig. 3.

   Fig. 3

   

   View largeDownload slide

   Stress from strain gage measurements versus stress from assumed value of K

   Fig. 3

   

   View largeDownload slide

   Stress from strain gage measurements versus stress from assumed value of K

   Close modal

2. A value of K between 0.05 and 0.50 is assumed.

3. Equations (12) through (14) are solved for each welded outer layer and the resultant curve based on an assumed value of K is plotted as shown in Fig. 3.

4. If the curve from the equations does not match the actual data, then a new value of K is assumed and the procedure is continued until a satisfactory curve is obtained.

5. Once the value of K is established, then the stress distribution in the various layers due to the wrapping process is determined from Eqs. (12) through (14). Figure 4 shows a typical stress distribution in the layered shell due to wrapping stress.

   Fig. 4

   

   View largeDownload slide

   Wrapping stress distribution through thickness

   Fig. 4

   

   View largeDownload slide

   Wrapping stress distribution through thickness

   Close modal

The wrapping stress in the various layers is added to the stress in these layers due to prestressing the concrete outer shell and stress due to internal pressure to obtain the overall stress pattern in the shell.

The thickness of the outer concrete shell and the amount of external pressure exerted by the prestress wire are obtained by trial and error. Three design criteria must be satisfied in choosing the concrete thickness and amount of external pressure. The first is the compressive stress in the concrete due to prestressing and the compressive stress in the steel shell due to prestressing and wrapping must not exceed the allowable compressive stress of the concrete and steel, respectively. The second criterion is the tensile stress in the steel shell due to internal pressure must not exceed the allowable stress and the stress in the concrete must remain in compression. And the third criterion is the stress in the steel shell due to hydrostatic testing must remain within the allowable stress and any tension in the concrete shell needs to be resisted by reinforcing bars.

where δ_sopi is the deflection of steel cylinder at outside surface due to internal pressure, δ_sopf is the deflection of steel cylinder at outside surface due to interface pressure, δ_cipf is the deflection of concrete cylinder at inside surface due to interface pressure, δ_cipo = deflection of concrete cylinder at inside surface due to outside pressure.

The above equations are used repeatedly as the thickness of the concrete and the external pressure are adjusted to meet the design conditions. When the thickness of the concrete and the external pressure are found satisfactory, then the number of prestress layers is calculated as shown next.

The external pressure P_o calculated above is provided by the prestress wires. The wires are placed spirally with no space between them. For ease of calculations, an effective thickness is determined by smearing the wire area over 1 in. of shell length to come up with an equivalent thickness per inch of length

where R_o =outside radius of the concrete shell for the first layer of wires and R_o = outside radius of the concrete shell plus t_w for the second layer of wires, etc.

The equations for calculating the stress in the concrete and steel inner shell due to placing the first layer of prestress wires are given below. When the second layer of prestress wire is placed, a second set of calculations is performed. However, the stress in the steel and concrete shells imposed by the first set of wires is now reduced due to the added second layer, which reduces the tensile stress in the first layer. This reduction has to be accounted for as the process proceeds from one prestress layer to the other.

The pertinent equations needed for the analysis are based on Eqs. (15), (18), and (19) and the details are shown in Fig. 6. The derivation of the equations is shown below.

where

and where, $λ=[3(1−μh2)(Rh/th)2]0.25$

where

Once the unknown forces are calculated from Eq. (39), the stresses are then determined from the following equations:

Steel head

Concrete shell

A finite element analysis of the total vessel is necessary in order to ascertain the stress levels because various loading conditions at various locations are within the allowable stress.

Hydrogen permeation in carbon steel is a major concern at high pressures. It can cause accelerated crack growth and possible embrittlement. Accordingly, stainless steel liners are used as a barrier. An experiment was conducted at ORNL to measure the effect of stainless steel on permeation [3,4]. Figure 8 shows a four-plate setup immersed in a hydrogen bath. Figure 9 shows the permeation rate of hydrogen in stainless steel versus carbon steel. It shows a permeation rate of carbon steel to be about ten times higher than that of stainless steel. Accordingly, a venting system is needed to dispose of the permeated hydrogen in the heads and shell to the atmosphere. This is accomplished by providing a lose type 304 stainless steel liner as a barrier. Hydrogen permeating through the stainless steel liner is vented to the atmosphere through vent holes in the heads and shell. Venting through the circumferential head-to-shell welds is more problematic. However, this can be circumvented by using a thin chill bar made of copper or titanium between the weld in the liner and weld in the steel as shown in Fig. 10.

The vent holes in the layered steel shell shown in Fig. 10 are covered by the outer concrete shell. In order to provide a path for the hydrogen to vent to the atmosphere, a grooved steel shroud shown in Fig. 11 is placed on the outside surface of the steel layered shell and underneath the concrete. The grooves fit over the holes to channel the hydrogen toward the heads, which are not covered by the concrete.

A full-scale vertical steel–concrete pressure vessel was designed and fabricated in accordance with the procedure outlined above. The design pressure was 6250 psi (43.1 MPa). The inside diameter of the vessel was 47 in. (1190 mm) and the length of the steel layered shell was 79 in. (2007 mm). A 20 in. (508 mm) manway was installed in the top head. A skirt length of 53.5 in. (1360 mm) was attached to the bottom head as shown in Fig. 12.

The heads were made of SA-537 CL 2 material with a tensile strength of 75 ksi (517 MPa) and minimum thickness of 2.48 in. (63 mm) with a type 304 inner stainless steel liner. The steel-layered shell was made of SA-724 material with a tensile strength of 95 ksi (655 MPa) with a type 304 inner stainless steel liner. It consisted of four layers. The thickness of each layer was 0.583 in. (14.8 mm) thick. All pressure components, with the exception of the steel layered shell, were designed to 6250 psi (43.1 MPa). The steel-layered shell was designed to 3620 psi (25.0 MPa) pressure. The steel vessel was hydrotested at 5180 psi (35.7 MPa) and stamped in accordance with the ASME Section VIII, Division 2.

The outer concrete shell was constructed in accordance with ACI 318 and ACI 372R and consisted of high strength concrete with a compressive stress, f′_c, of 9000 psi (62.1 MPa). The thickness was 11.2 in. (284 mm). The prestress wires consisted of 5 layers and were ASTM A-648 Class III with a diameter of 0.192 in. (5 mm) and a tensile strength of 252 ksi (1737 MPa). The reinforcing bar cages were ASTM A-706 with a tensile strength of 80 ksi (552 MPa). Details of construction are shown in Fig. 13.

The allowable stress in the prestress wires is taken as 176,400 psi (1216 MPa). The relaxation stress is taken as 35,000 psi (241 MPa). Hence, the design stress used in the analysis is 141,400 psi (975 MPa).

A thin layer of mortar was added to the outside of the prestress wires for protection as shown in Fig. 14.

A finite element analysis was performed on the heads and shell. All critical stress values were within the allowable stress levels. Figure 15 shows a typical output for the analysis.

The completed vessel was hydrotested at 8940 psi (61.6 MPa). The vessel, Fig. 16, was then cycled with hydrogen to prove its integrity.

The research was sponsored by the U.S. Department of Energy with Oak Ridge National Laboratory (ORNL), managed and operated by UT-Battelle, LLC. Programmatic direction was provided by the Fuel Cell Technologies Office. Mr. Nico Bouwkamp of the CaFCP & Frontier Energy was very helpful in showing the authors the existing infrastructure and various kinds of fueling stations in the Los Angeles, California, area. The cost analysis of various configuration options was done by Mr. Mike Kelly of Irvine, California. The initial analysis of the concrete shell including Finite Element analysis was performed by Dr. Fariborz Vossoughi at the Ben Gerwick Corporation in Oakland, California. Final details of reinforcement and prestressing of the concrete were calculated by Mr. Basil Kattula of Kent, Washington. A comprehensive Finite Element Analysis was also conducted by Jian Chen at ORNL. The layered steel vessel was fabricated by Kobelco in Kobe, Japan, with Mr. Susumu Terada contributing substantially to the various phases of fabrication. Strain gage instrumentation and data acquisition during concrete placement, hydrotesting, and hydrogen cycling was conducted by Dr. Fei Ren of Temple University in Baltimore, Maryland. The prestressed and reinforced concrete shell was fabricated by the Thompson Pipe Group in Grand Prairie, Texas, with Mr. Sam Arnaout's technical assistance. Final hydrotesting and hydrogen cycling was conducted at the Harris Thermal Transfer Products in Newberg, Oregon, under the direction of Mr. Jim Nylander.

This manuscript has been co-authored by UT-Battelle LLC, under Contract No. DE-AC0500OR22725, with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan.²

• U.S. Department of Energy (Contract No. DE-AC05-00OR22725; Funder ID: 10.13039/100000015).

d =

diameter of prestress wire

D =

diameter of vessel

E =

modulus of elasticity

E_c =

modulus of elasticity of concrete

E_h =

modulus of elasticity of head

E_s =

modulus of elasticity of shell

K =

weld shrink factor obtained experimentally

L =

length of shell

n =

number of weld seams in a given layer

N =

number of layers of prestress wire

P =

design pressure

P_f =

interface pressure

P_i =

internal pressure

P_o =

external pressure

R =

radius at desired location

R_c =

radius of concrete shell (average of inside and outside radii)

R_h =

radius of the head (average of inside and outside radii)

R_i =

inside radius

R_o =

outside radius

R_s =

radius of steel shell (average of inside and outside radii)

S =

allowable stress

S_w =

allowable stress of prestress wire

t =

thickness

t_c =

thickness of concrete shell

t_h =

thickness of head

t_i =

thickness of all layers underneath t_o

t_o =

thickness of outer layer being shrunk

t_s =

thickness of steel shell

t_w =

equivalent thickness of prestress wire

w =

width of weld seam

δ =

radial deflection

δ_c =

radial deflection of concrete shell

δ_h =

radial deflection of head

δ_i =

radial deflection of the inside surface of a cylinder

Δ_i =

radial deflection of inner layer t_i due to interface pressure P_f

δ_o =

radial deflection of the outside surface of a cylinder

Δ_o =

radial deflection of outer layer t_o due to interface pressure P_f

δ_s =

radial deflection of steel shell

δ_w =

radial deflection of layer t_o due to weld shrinkage

Δ_w =

weld shrinkage

ε_L =

longitudinal strain

ε_θ =

circumferential strain

θ_c =

rotation of concrete shell

θ_h =

rotation of head

θ_s =

rotation of steel shell

μ =

Poisson's ratio

μ_c =

Poisson's ratio of concrete shell material

μ_h =

Poisson's ratio of head material

μ_s =

Poisson's ratio of steel shell material

σ_L =

longitudinal stress

σ_θ =

circumferential (hoop) stress

σ_θc =

circumferential stress in concrete

σ_θs =

circumferential stress in steel