Transactions, SMiRT-25

Charlotte, NC, USA, August 4-9, 2019

Division VI

COMPARISON STUDY BETWEEN 2D REINFORCED CONCRETE

ELEMENT DESIGN METHODOLOGIES USING AUTOMATED DESIGN
SOFTWARE FOR NUCLEAR STRUCTURES
James Phillips1, Ahmed Alshawi2, Yasuo Nitta3, Takaaki Tsukada4
1
Principal Engineer, Mott MacDonald, Altrincham, UK ([email protected])
2
Senior Engineer, Mott MacDonald, Altrincham, UK ([email protected])
3
Senior Engineer, Shimizu Corporation, Tokyo, Japan ([email protected])
4
Structural Engineer, Shimizu Corporation, Tokyo, Japan ([email protected])

ABSTRACT

The objective of this paper is to evaluate and compare three different common design methodologies for
the design of reinforced concrete 2D shell elements (slabs and walls) in nuclear safety related structures.

Automated design software, that has been developed in accordance with these three design
methodologies, is utilised to compare utilisation ratios found for the three different methods for a realistic
example. The data used for the example is taken from a UK safety related structure finite element analysis
model.

The three methodologies compared within this paper are:

1. 2D Allowable Stress Method – This approach assesses the stresses across the section and
compares them to allowable stresses (Nakamura, 1997).
2. Clark-Nielsen Method (also known as the sandwich method) – This approach utilises a sandwich
model to derive forces in two outer reinforcement planes and then resolves the forces in those
planes using formulae derived by Nielsen for thin shell elements (Cope and Clark, 1984).
3. Direct compliance to ACI349M-13 (ACI Committee 349, 2017) – This assesses the reinforcement
requirements utilising load-moment integration graphs and in-plane shear checks as per ACI
349M-13 and is outlined within the ACI318-08 PCA notes (Kamara, Novak and Rabbat, 2008).

Comparison of the results found that there are typically small differences between the utilisation
ratios of the three methods. The results show a strong linear relationship between results with trend lines
close to a 1:1 ratio (within 6% for all the comparisons). The results also highlighted benefits in considering
the angle of principle stress where one axis is in compression and the other in tension.

The conclusions from this comparison indicate that the selection of design method should not have
a fundamental effect on the final design outcome with regards to total reinforcement and utilisation ratios.

INTRODUCTION

Within the nuclear industry, finite element analysis (FEA) is commonly adopted to support the design of
nuclear structures due to their scale and complexity. Where reinforced concrete structures consist of
diaphragm slabs and shear walls, these elements are typically modelled as 2D shell elements within the
analysis. Once analysed, eight output forces and moments are generated for each element (three in-plane
forces, three out-of-plane moments and two out-of-plane forces). These FEA models typically consist of
hundreds of thousands of shell elements, which are each assessed for over 100 load cases.
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

To design for these shell element forces there are a number of different design methodologies and
code approaches that are adopted within the international nuclear industry. The choice of methodology
and/or design code depends upon national regulatory requirements, Client requirements, and/or Designer
preference.

The objective of this paper is to evaluate three different design methodologies for the design of
reinforced concrete 2D shell elements (slabs and walls), that have previously been adopted internationally
on different nuclear safety related projects. For each of these design methodologies, automated design
software is to be utilised to compare the reinforcement utilisations from the assessment of an example
seismic analysis of a typical nuclear safety related structure.

The paper sets out differences and similarities between the different design philosophies,
methodologies, assumptions and limitations and also compares and discusses the reinforcement utilisation
ratio results obtained for each method.

DESIGN METHODOLOGIES

The three different design methodologies compared are:

1. 2D Allowable Stress Method – Utilises an allowable stress method based on engineering first
principles to assess stresses in reinforcement and concrete for a set structural layout by
determining linear stress and strain diagrams in two directions considering the angle of principal
strain.

2. Clark-Nielsen Method (also known as the sandwich method) –Utilises a sandwich model to
derive forces in two planes and then resolves the forces in those planes using formulas derived by
Nielsen for thin shell elements. This approach is similar to those outlined in Eurocode 2 (CEN/TC
250, 2010), in-built in some software and outlined in other publications.

3. Direct compliance to ACI349M-13 – This assesses the reinforcement requirements for the two
planes separately utilising load-moment integration graphs and in-plane shear checks as per ACI
349M-13 and as outlined within the ACI318-08 PCA notes. Note: The design methods within the
ACI 349M-13 and ACI 318M-08 document revisions are not specifically written for use with
element by element design and therefore design assumptions have historically been made when
adopting this approach.

A comparison of the design methodology, assumptions, limitations and benefits is provided within
Table 1.

It has previously been expected by Designers that designing to the ACI 349M-13 method should
produce slightly more economical results than the 2D Allowable Stress method. This is due to the ACI
349M-13 method adopting a design strength approach assuming yielding of the concrete up to a maximum
usable strain. In addition, it has been understood by Designers that the Clark-Nielsen Method should
provide a realistic but conservative design, compared to other more accurate methods, due to its simplified
approach. These expectations are evaluated further in this paper, as well as the extent of any differences.

AUTOMATED DESIGN SOFTWARE

Designers are increasingly adopting automated design processes to reduce project costs and programme
and to help manage large scale output results from FEA models. For each of the design methodologies
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

outlined above, Shimizu Corporation with the support of Mott MacDonald have developed automated
design software to enable the fast and economic design of reinforced concrete 2D shell elements. These
three pieces of software are as follows:

1. SSDP-2D – which adopts an allowable stress method

2. SSDP-CN – which adopts a Clark-Nielsen method
3. SSDP-WS – which adopts an ACI 349M-13 design method.

Each of the automated design software allow for assessment of each element and each load
combination on a case by case basis, which can lead to reductions in reinforcement requirements and
reduced design time over existing manual designs. Each of the three pieces of software have been through
an extensive validation and verification process, the limitations and capabilities of each software are briefly
discussed in Table 1.

As part of the research for this paper, the three pieces of automated design software have been used
to produce reinforcement utilisations ratios (URs) for results from a UK nuclear safety related structure.
The outputs to be compared from the software are the maximum utilisation ratios for each reinforcement
direction (i.e. maximum of top and bottom reinforcement URs).

STRUCTURE FOR COMPARISON

To enable the comparison of the three different methodologies in a realistic scenario, a typical nuclear safety
related structure is assessed. For this assessment, inputs and results, developed from an FEA, for a
previously designed UK nuclear safety related structure are used.

The structure assessed is a large reinforced concrete shell structure consisting of diaphragm slabs
and shear walls. The structure is approximately 70m long x 20m wide x 25m tall and supports a number of
large plant loadings.

For this study, one diaphragm floor slab and one of the main shear walls are evaluated, providing
a total of 4396 elements assessed in both axes. See Figure 1 for example plots showing the configurations.
The results from the ‘Dead + Live – Seismic’ (D+L-E) and ‘Dead + Seismic’ (D+E) load combinations are
considered, to ensure a mix of compressive and tensile loaded members within the results. Note: All seismic
forces are assigned the same sign for the example assessment with +ve being tensile and -ve compressive.

The input strength reduction factors and material properties are the same for all three design
methods for fair comparison. No slenderness effects are considered. The results from the ACI 349M-13
method are based upon those determined at a construction joint, as this method more directly correlates to
the 2D Allowable Stress and Clark-Nielsen methods.

Figure 1. Left - Plan of slab assessed (Mx force plot). Right - Elevation of wall assessed (Txy force plot)
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

Table 1: Comparison of different 2D reinforced concrete element design methods

2D Allowable Stress Method (SSDP-2D) Clark-Nielsen Method (SSDP-CN) ACI 349M-13 method (SSDP-WS)
The SSDP-2D design is based upon several The design basis for SSDP-CN is an approach known The design basis for SSDP-WS is based on the
documented engineering first principles, with commonly in the UK as Clark-Nielsen (or the sandwich requirements of Chapter 10 of ACI 349M-13 for
regards to equilibrium of forces, the stress-strain method), which has been documented in numerous papers, combined flexure and axial loads and the
relationships across a section, and the angle of text books and design codes. requirements for in-plane shear in Section 11.6 and
principal stress/strain in the plane of the element. 11.9.
The Clark-Nielsen design method assumes that the RC
SSDP-2D determines stresses and strains within the shell element is divided in to three membrane layers (two Flexure and Axial Loads (Chapter 10)
reinforcement and concrete by resolving equations reinforced outer layers and an inner unreinforced concrete Reinforcement requirements for the flexure and axial
based on equilibrium of external forces and internal core), as shown below. loads are determined using P-M interaction diagrams,
section forces. These equilibrium equations are which are developed in accordance the ACI 318-08
developed considering the external forces and the PCA notes. This is determined for the vertical and
internal section stress and strains distribution in 3 horizontal directions separately, for each finite
dimensions. A linear relationship between the element with the relevant forces and moments as
stress and strain for concrete and steelwork is follows:
assumed (i.e. the materials remain perfectly elastic).
Vertical Direction: Ny – (My + Mxy) interaction curve
The software initially determines the angle of Horizontal Direction: Nx – (Mx + Mxy) interaction
principle stress/strain, by iteration, to determine the curve
primary and secondary crack directions (see image
below). It then iterates to determine the crack The interaction diagrams are developed by
depths in these two directions. From this, the determining a number of different points along the
software is then able to resolve the external and interaction diagram. The strain diagrams for the
internal section forces to determine the different parts of the load-moment interaction graph
reinforcement and concrete stresses. are shown below.

The design forces in each of these layers are determined by

resolving the axial forces and moments about the centre
lines of the outer layers including both in-plane shear and
torsional moments.
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

2D Allowable Stress Method (SSDP-2D) Clark-Nielsen Method (SSDP-CN) ACI 349M-13 method (SSDP-WS)
Equilibrium equations: A simple example is shown below for a single direction,
along with formulae for each direction.

Where ‘l’ and ‘m’ are the direction cosines of the

anisotropic coordinate system (1-2 axes) in the X-Y
coordinate system.

Once the forces within the two membrane layers are Account is taken of the reinforcement in the
determined (Nx, Ny, Nxy), the reinforcement requirements compression zone, with the maximum compressive
and concrete stresses in the membrane are determined resistance increasing with increasing reinforcement.
using formulae originally developed by Nielsen.
The capacity at each point on the curve is determined
These consider: considering equilibrium of forces, ensuring that the
1. Whether the resultant x and y design (NxDes, NyDes concrete and steel strain capacities are equal or less
– see below) forces are in tension or compression than code defined limits and assuming a rectangular
2. The optimum principle strain direction (i.e. to give a concrete stress distribution.
minimum total reinforcement).
Once the initial curve is developed it is amended to
Case 1 – reinforcement required in both directions, consider the strength reduction factors as per ACI
NxM and NyM > -|NxyM| 349M-13.
NxDes = NxM + |NxyM|,
NyDes = NyM + |NxyM|, The above approach can be utilised to determine
Fc =-2 |NxyM| reinforcement requirements via an iterative approach.
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

2D Allowable Stress Method (SSDP-2D) Clark-Nielsen Method (SSDP-CN) ACI 349M-13 method (SSDP-WS)
Case 2 – reinforcement required in X-direction only, In-plane Shear (Sections 11.6 and 11.9)
NxM > -|NxyM| and NyM ≤ -|NxyM| The reinforcement requirements to resist in plane
NxDes = NxM - NxyM² / NyM shear forces, Nxy, are dependent on whether the
NyDes = 0 section is located at a construction joint. For non-
Fc =NyM + NxyM² / NyM construction joints the reinforcement is determined
using ACI 349M-13 equations 11-29 and 11-30. For
Case 3 – reinforcement required in Y-direction only, construction joints the reinforcement is determined
NyM > -|NxyM| and NxM ≤ -|NxyM| using ACI 349M-13 equation 11-25.
As above but reversed
The horizontal and vertical reinforcement are
Case 4 - no reinforcement required, determined utilising the same approach.
NxM and NyM ≤ 0 and NxMNyM > NxyM²
NxDes = 0 Where compressive loads occur across an element the
NyDes = 0 in-plane shear loads, and resultant in-plane shear
Fc = 0.5(NxM + NyM) – 0.5[(NxM-NyM)² + 4NxyM²]0.5 reinforcement can be reduced, as outlined in the ACI
349M-13, cl. 11.6.7.
NxDes and NyDes forces are used to determine the X-
direction and Y-direction reinforcement. The reinforcement requirements to resist in-plane
Fc is used to check the concrete stress. shear forces, Nxy, are assessed separately to the axial-
flexure assessment and then the reinforcement
requirements for the two assessments are combined.
Software Capabilities and limitations: Software Capabilities and limitations: Software Capabilities and limitations:
• Calculates stress within reinforcement and • Calculates force within reinforcement and concrete to • Calculates required reinforcement areas in
concrete for comparison to allowable stresses. calculate required area of reinforcement. accordance with ACI 349M-13.
• Considers direction of principle stress and strain. • Considers direction of principle stress and strain that • Considers x and y axis separately.
• Requires reinforcement area to be input. results in optimum reinforcement. • Can determine area of reinforcement required or
• Unsymmetrical reinforcement can be assessed. • Can determine area of reinforcement required or utilisation ratios for input reinforcement areas
• Does not account for slenderness within software. utilisation ratios for input reinforcement areas • Only symmetrical reinforcement can be assessed
• Shear friction assessment for in-plane shear at • Unsymmetrical reinforcement can be assessed. • Accounts for slenderness
construction joints inherent within design • Does not account for slenderness within software. • Designs for requirements at non-construction and
methodology. • Reinforcement only acts in tension. construction joints in accordance with ACI 349M-
• No consideration given for concrete shear • Software not suited for thin sections where cover > h/4. 13.
capacity at non-construction joint. • Shear friction assessment for in-plane shear at
construction joints inherent within design methodology.
• No consideration given for concrete shear capacity at
non-construction joint.
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

RESULT COMPARISONS

The utilisation ratios for the three different methods are compared in Figure 2 for the slab and Figure 3 for
the wall. In each of these figures, the URs from the different design methods are plotted against each other
for each element; with all three methods compared to each other across three plots. On each figure a Y=X
line is plotted in red to allow for ease of assessment of the results. The results for the reinforcement for both
reinforcement axis (X & Y) have been placed on the same plot. The comparisons are separated for the two
load combinations, D+L-E (typically compression) and D+E (typically tension) to assess the effect on
tension and compression design cases.

Comparisons between concrete stresses are not presented in this paper, as the ACI 349M-13 method
assumes concrete is stressed to max allowable and therefore the comparison isn’t informative. However, a
review by the author of the concrete stresses determined for the 2D Allowable Stress method and the Clark-
Nielsen method indicated that they were well within the allowable stresses with low UR values.

For the D+E load cases, trendlines have been added to the figures in blue to determine the average
differences between URs for the different methods. Trendlines are not added for the D+L-E load case
(seismic compressive load case), as there are a large number of low URs for this load case which distort the
trendline, particularly where reinforcement is not required.

The results show a very strong correlation between the different methods, with the exception of
some instances where low URs occur (D+L-E cases), because of differences in the design methods where
compressive loads are present.

From review of the D+E load case results (tension seismic load), it can be seen that there is a strong
linear relationship between the results for the different methods. The linear trendlines plotted indicate that
the average trend of the results are within 6% percent of a 1:1 relationship for all comparisons.

From review of the D+L-E Seismic load case results, it can be seen that the majority of results give
low UR values, due to the load combination typically resulting in beneficial compression across the section.
For the higher UR values (>0.4) the 2D Allowable Stress method and Clark-Nielsen method resulted in
lower URs than the ACI 349M-13 method. All of these higher UR values occurred where the axial forces
in one direction remained in tension and the other axis axial forces were in compression. The lower URs
for the 2D Allowable Stress method and Clark-Nielsen method are due to these design methodologies
considering the direction of principle stress, which leads to reductions in tensile reinforcement where the
other axis is in compression. The automated design software for the ACI 349M-13 method does not consider
the direction of principle stress and therefore its benefit isn't realised in these instances.

In-built within the SSDP-WS software, which designs to the ACI 349M-13 methodology is the
ability to assess sections at construction joints and at non-construction joints. In the comparisons in Figures
2 and 3, the reinforcement was assessed at construction joints, as this approach for considering in-plane
shear loads is more aligned to those within the 2D Allowable Stress method and Clark-Nielsen method. An
initial comparison of the URs for the construction joint and for the non-construction joint indicated some
large reductions in URs for the non-construction joint design. Reviewing these potential savings is outside
of the scope of this paper, but these will be investigated in future studies.

Legend for Figure 2 and Figure 3

1:1 Utilisation ratio
Trend line between methodologies including equation
Utilisation ratios for elements
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

2a. Slab D+E, SSDP-2D to SSDP-WS 2b. Slab D+L-E, SSDP-2D to SSDP-WS
1.2 1.2

1 1
y = 1.0055x
0.8 0.8

SSDP-WS
SSDP-WS

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
SSDP-2D SSDP-2D

2c. Slab D+E, SSDP-2D to SSDP-CN 2d. Slab D+L-E, SSDP-2D to SSDP-CN
1.2 1.2

1 y = 1.0149x 1

0.8 0.8
SSDP-CN
SSDP-CN

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
SSDP-2D SSDP-2D

2e. Slab D+E, SSDP-CN to SSDP-WS 2f. Slab D+L-E, SSDP-CN to SSDP-WS
1.2 1.2

1 y = 0.9906x 1

0.8 0.8
SSDP-WS
SSDP-WS

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
SSDP-CN SSDP-CN

Figure 2. Comparison of Utilisation Ratios for different design methodologies for diaphragm slab
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

3a. Wall D+E, SSDP-2D to SSDP-WS 3b. Wall D+L-E, SSDP-2D to SSDP-WS
1.4 1.2
y = 0.958x
1.2 1
1
0.8
SSDP-WS

SSDP-WS
0.8
0.6
0.6
0.4
0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2
SSDP-2D SSDP-2D

3c. Wall D+E, SSDP-2D to SSDP-CN 3d. Wall D+L-E, SSDP-2D to SSDP-CN
1.4 1.2
y = 1.0134x
1.2
1
1
0.8
SSDP-CN
SSDP-CN

0.8
0.6
0.6
0.4
0.4
0.2
0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2
SSDP-2D SSDP-2D

3e. Wall D+E, SSDP-CN to SSDP-WS 3f. Wall D+L-E, SSDP-CN to SSDP-WS
1.4 1.2
y = 0.946x
1.2 1
1
0.8
SSDP-WS

SSDP-WS

0.8
0.6
0.6
0.4
0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2
SSDP-CN SSDP-CN
Figure 3. Comparison of Utilisation Ratios for different design methodologies for shear wall
 25th Conference on Structural Mechanics in Reactor Technology
Charlotte, NC, USA, August 4-9, 2019
Division VI

CONCLUSIONS & DISCUSSIONS

A comparison has been made between three different reinforced concrete shell element design methods that
are commonly adopted internationally in the nuclear industry. The three methods all adopt different
engineering principles to assess reinforcement requirements and utilisation ratios for a given section and
load configuration. From the formulae for each method alone it is difficult to identify if any one given
method is more economical or whether the different methods are directly comparable.

The comparison of the results indicates that the results for tension load combination (D+E) are very
similar for all three design methods with the plotted trendlines showing strong correlation (y=1.005x,
y=1.015x, y=0.991x, y=0.958x, y=1.013x & y =0.946x).

The majority of the results for the compression load combination (D+L-E) resulted in very low UR
values (<0.2), due to the sections being in compression. Higher UR values (>0.4) still occurred for some
elements, due to the presence of tension in only one direction. For these higher UR cases, the 2D Allowable
Stress method and Clark-Nielsen method produced lower URs than the ACI 349M-13 method, due to the
design methods considering beneficial effects of the direction of principle stress.

Prior to conducting the assessment, it was expected that the ACI 349M-13 method would produce
the most economical results due to it adopting a strength design method assuming yielded concrete and
maximum usable strains. The results indicate however, that the approach yields similar results to those from
the allowable stress method. The Clark-Nielsen Method typically produced the most conservative design,
as anticipated, but the extent of these conservatisms is very low and should not affect design.

The final choice of method used for design will often be dependent on the Client, local regulatory
requirements and Designer preferences, however, the works within this paper indicate that the selection of
design method should not have a fundamental effect on the final design outcome with regards to total
reinforcement and utilisation ratios. An exception to this may be for elements and load cases where one
axis is in tension and the other is in compression; for which there was found to be a reduction in URs when
using the 2D Allowable Stress method or the Clark-Nielsen method due to the beneficial effects of
considering the angle of principle stress.

Comparison of utilisation ratios at construction joints and non-construction joints indicated large
potential savings at non-construction joints. A further review into the extent of these savings and the impact
on design and practical implications for construction is to be investigated in future studies.

REFERENCES

ACI Committee 318 (2008). Building Code Requirements for Structural Concrete (ACI318M-08) and
Commentary, American Concrete Institute, Farmington Hills, MI, U.S.A.
ACI Committee 349 (2017). Code Requirements for Nuclear Safety Related Structures (ACI349M-13) and
Commentary, American Concrete Institute, Farmington Hills, MI, U.S.A.
M.E. Kamara, L.C. Novak , B.G. Rabbat (2008). Notes on ACI 318-08 Building Code Requirements for
Structural Concrete with Design Applications, Portland Cement Association, Skokie, Illinois, U.S.A
M.P.Nielsen, (1984). Limit Analysis and Concrete Plasticity 1st Edition, Prentice Hall.
M. Nakamura (1997), “Stress Evaluation of RC Shell Elements based on Three-dimensional Equilibrium”,
the SMiRT14 Conference Proceedings, Division H, Lyon, France, August 17-22
R.J. Cope and L.A. Clark (1984), Concrete Slabs Analysis and Design, Elsevier Applied Science Publishers
Ltd.
Technical Committee CEN/TC 250 (2010). BS EN 1992-2:2005, Eurocode 2: Design of Concrete
Structures – Part 2: Concrete bridges – Design and detailing rules, British Standard Institution.