Structural Characteristics of Reinforced Palm Kernel Shell Concrete Deep Beams

This paper evaluates the structural characteristics of deep beams made from reinforced palm kernel shell concrete (PKSC) and normal weight concrete (NWC). Twelve PKSC and NWC deep beam samples, with and without shear reinforcement were tested under three-point loading and their structural behavior studied. The ultimate shear strength of PKSC beams increased with a decrease in the shear span-todepth ratio. Post diagonal cracking shear resistance is greater in PKSC deep beams than beams of normal weight concrete. The shear capacity of the PKSC and NWC deep beams were assessed to be un-conservative using ACI 318-99, ACI 318-05, Euro code (EC) 2 and a kinematic model, when compared with the experimental results. Nonetheless, this necessitated the development of a calibration procedure to correct the bias inherent in these models. Calibrated shear strength models revealed the compressive strength and the ratio of the shear span-to-total depth as significant influential parameters for correcting the inherent bias in the original deterministic shear strength models. The calibrated functional model of ACI-318-99 may produce conservative predictions, given this limited number of test specimens. Therefore future studies should investigate the reliability of the calibrated models, and quantifying the uncertainties in the estimated coefficients of parameters, using a much larger representative dataset.


Introduction
Concrete has a widespread application in the construction industry.The increased demand for concrete as a construction material, has yielded the investigation into unconventional building materials such as steel milled from scrap metals [1][2][3], bamboo reinforcement in concrete [4][5][6], phyllite aggregate waste in concrete [7], palm kernel shell aggregate in concrete [8,9].Conventionally, rocks of igneous, metamorphic and sedimentary origins such as granite, basalt, flint, limestone etc. [10,11] have been used to produce coarse aggregates for concrete production over the years.Nonetheless, there has been increased use of coarse aggregates, which serve as economic filler in concrete, for construction purposes.This practice is leading to over-exploitation of natural rock resources in the environment and a depletion of sources of coarse aggregates.
There is therefore the need to find alternative coarse aggregate resources to replace natural ones.Research conducted on various solid waste materials; such as granulated coal ash, blast furnace slag, fiberglass waste materials, granulated plastics, sintered sludge pellets, phyllites from mining waste, ceramic waste and recycled concrete; as coarse aggregate substitutes has proved very successful [7,12,13].Most of these waste materials constitute artificial light weight aggregates [14] and their introduction to replace conventional aggregates for concrete production in some developed countries has been quite beneficial to the development of infrastructure [15].Since the major characteristic properties of coarse aggregates impact directly on the resulting concrete properties such as compressive strength, shear strength, ductility behaviour and durability, it is important to evaluate the performance of alternative aggregate materials before accepted for general use.Palm Kernel Shell (PKS), a waste product from palm kernel oil production, is one such material that is receiving attention in recent times for use in the concrete industry due to its superior performance characteristics as coarse aggregate in lightweight concrete production [8].
One such use for PKS aggregate concrete is in the construction of RC beams.RC beams are classified into slender and deep beams [16].Some previous studies conducted on PKSC slender beams have shown that it has satisfactory performance characteristics in compression, shear, flexure, ductility, aggregate interlock and bond with steel reinforcement [8,14].Hence, can be employed, especially in low-cost housing construction and also for use in earthquake prone areas due to its low weight advantage and ductility characteristics.On the other hand, RC beams where the shear span is less than twice the member depth are characterized by arch action rather than beam action, and are known as deep beams [17,18].More so, RC beams are known to exhibit low shear resistance when the depth of the beam increases [19,20].The conventional Euler-Bernoulli beam is usually not applicable to characterizing the shear behavior of such members, and as such, design concepts like strut and tie models, seems preferred.Recently, Tasenhod and Teerawong [21] developed an improved strut and tie model based on ACI 318-11.Kassem [22] adopted a fixedangled softened truss model in the development of shear strength model for concrete deep beams, yielding reliable results than shear provisions as specified in the ACI code and Eurocode.Chou [23] presented a nature-inspired metaheuristic regression method for predicting shear strength of RC deep beams.It is worth noting that the developed shear strength models in these studies, employs a data set of conventional RC deep beams.
Virtually no research work has focused predicting the shear strength of PKSC deep beams.PKSC, as a light weight material for construction, could be a suitable material for the construction of RC deep beams as transfer girders in highrise buildings designed to withstand seismic and other inertia forces due to their weight advantage and high ductility characteristics.The importance of the use of lightweight concrete in high-rise construction is highlighted by Metha and Monteiro [24], especially where the low bearing capacity of the soil was problematic.In one case, a lightweight concrete mix was used in the construction of a 52-storey building instead of 35-storey had it been a normal concrete building.
Deep beams usually fail through shearing action which tends to be more sudden and brittle [17,25].However, the accurate prediction of the shear strength of reinforced concrete deep beams is vital in engineering design and management [23].Nonetheless, several of the developed shear strength models for RC deep beams either overestimate [21] or under estimate [26,27] experimental observations.More so, studies that have evaluated the adequacy of codebased shear provisions have either reported relatively lower correlation coefficient [23] or larger coefficient of variation [26] in comparison with developed models.The adequacy of these models is highly reliant on the range of variables considered as well as the size of the database employed [28].Hence it is becomes necessary to develop explicit shear strength model for PKSC deep beams.This study investigated the shear strength of 12 reinforced PKSC and Normal Concrete (NC) deep beams.The main variables considered in the study were shear span-to-effective depth ratio, vertical shear reinforcement ratio, and type of concrete.The shear capacity predictability of 3 codes of practice; ACI 318-99 (empirical formulae based), and ACI 318-05 and EC 2 (both based on strut-and-tie models) and one kinematic model were evaluated for such RC deep beams.A calibration procedure is proposed to provide reliable and consistent shear strength estimate for engineering design.

Materials and Method
According to Shetty [29], lightweight concrete mix design is usually established by trial mixes.Based on this method, a grade 30 concrete labeled as PKSC30 was designed, cast and tested for PKSC deep beams.Ordinary Portland cement of Grade 42.5 was used to achieve this target 28-day average compressive strength.The ratio of sand to cement and PKS to cement were 1.8 and 0.8 respectively based on a study conducted by [8].Another concrete mix was produced from normal weight aggregate (granite) labelled NWC30 which was also designed with a target compressive strength of 30 MPa.Table 1 shows the mix proportions for the two types of concrete that were produced for analysis in the study.A total of 12 concrete beams, categorized into four groups were tested to ultimate failure.For each group the variable investigated is the shear span-to-depth ratio.Six concrete beams (Group 1 and 2) were without shear reinforcement.Group 1 and 3 were made of PKSC deep beams (see Table 2).Each group consist of three beams with clear spans of 500, 815 and 1125 mm, each with a constant cross section of 150 × 350 mm.All beams were designed to fail in shear.
Sufficient reinforcement anchorage was provided near the supports by providing a 150mm overhang beyond the support center lines to avoid premature anchorage failure.The beams were reinforced with 1.0% mild steel longitudinal reinforcement with yield strength of 413 MPa and elongation of 22% and cast in timber molds.
Six concrete cubes (150 ×150 ×150 mm) with six companion concrete modulus of rupture beams (100 × 100 × 500 mm) were cast, cured in water and tested after 28 days.The main variables considered in the study for the RC beams were shear span-to-effective depth ratio (a/d), vertical shear reinforcement ratio (ρv) and type of concrete (PKSC or NWC).Table 2 provides details of each concrete beams and Figure 1 shows a typical beam setup ready for testing.In Table 2, the beams are labelled first with the concrete type ('P' for PKSC or 'N' for NC), followed by the shear spanto-depth ratio (1.0, 1.5, 2.0) and vertical shear reinforcement (S0 for beams without vertical shear reinforcement and S1 for beams having shear reinforcement).
The loading and measuring system consisted of incrementally applying a monotonic load of 2 KN/min till ultimate failure is observed for test setup.An I-section steel spreader was place on top of each beam, and with the help of two roller supports, these monotonically increasing loads were transmitted, as schematically presented in Figure 1.The first flexural crack load, first diagonal cracking and ultimate failure were recorded along with their respective mid-span deflections measured with a dial gauge beneath the soffit of the beam under test.The maximum crack width and crack height were also taken, and crack patterns were marked on tested beams during loading.

Table 2. Test Matrix
Where ln is the clear span measured between edges of both support plates; d is the effective depth of the beam; a/d is the shear span to depth ratio; ρb is the longitudinal bottom reinforcement ratio; ρv is the vertical shear reinforcement ratio and fcu is the cube compressive strength of concrete.Laboratory test results for the coarse aggregates are given in Table 3 for both PKS and granite.The results indicate as reported by others [8,30] that granite aggregates are superior to PKS in terms water absorption and density, which are parameters often linked to the durability and strength properties of concrete.The PKS however had a lower value of AIV compared to that of granite aggregates indicating that PKS aggregate concrete may possess a higher impact absorbance ability.The AIV is a measure of the relative resistance of aggregates to impact loads.
The 28-day compressive strength and modulus of rupture results were 33.70 and 1.19 MPa for the PKS and 42.93 and 2.88 MPa for the NWC samples respectively.The results of the concrete cube compressive strengths and those of the companion modulus of rupture beams are consistent with results from other researchers.

Load-Deflection and Structural Behavior
The load-deflection response of the PKSC and NC deep beams with and without shear reinforcement shows that as the a/d increased from 1.0 through 1.5 to 2.0, the failure load of the beams decrease (Figure 2).

Figure 2. Typical load-deflection curve for PKSC deep beams without shear reinforcement
The PKSC deep beams experienced larger deflections in most cases at failure compared to the NC deep beams, particular for beam with vertical shear reinforcement (see Table 4).This conforms to findings of Alengaram et al. [8] that emphasizes that PKSC slender beams exhibit higher deflections at failure and hence possess higher ductility characteristics at failure than similar NWC slender beams.The deep beam samples without shear reinforcement failed by strut failure through diagonal-splitting, which signified the splitting of the concrete struts formed between the loading and the support points with a loud explosion at failure.The failure mode for beams without vertical shear reinforcements (Figure 3) was primarily diagonal tensile splitting of concrete struts, whereas those with vertical reinforcement failed in shear compression (Figure 4).These failure modes were independent on the type of concrete type.For instance, beams in Group 1 and 2 observed the same mode failure (see Table 4).In terms of crack patterns, flexural cracks appeared first around mid-span of the beams, as loading was applied, and these were followed by diagonal cracks, which emanated from the tip of the support plate at the bottom of the beam towards the loading plate on top of the beam for both PKSC and NC deep beam samples.

Table 4. Summary of test results
Pmax is the maximum applied load; Vcr,exp is the diagonal cracking shear; Vu,exp is the ultimate shear; Δ is the midspan deflection; vnorm is the normalized shear stress and is computed as Vu,exp/(bdfcu); fcu is the cylindrical compressive strength in MPa; DS is diagonal splitting; SC is shear compression.Unlike the beams without shear reinforcement, the beams with vertical shear reinforcement experienced a more ductile failure with more number of cracks but less crack widths before failure (Figures 3 and 4).The PKSC deep beam samples experienced more number of cracks and wider crack widths at failure compared to the NC deep beams.Several flexural cracks appeared, as loading was increased until major diagonal strut cracks were observed between the loading plate and the support plates, which opened up gradually before failure occurred.
The shear strength of PKSC beams increased as a/d decreased (see Table 4).Nonetheless, the behavior of these beams after the formation of diagonal cracks is of prime importance.The parameter Vc,exp/Vu,exp is used herein to assess the extra load sustained after the initiation of the first diagonal crack (Vc,exp) to the ultimate failure (Vu,exp).The increase in shear strength above the diagonal cracking load ranged between 10 to 140 %.The post-diagonal cracking shear resistance is greater in PKS than NC beams (Vc,exp/Vu,exp), irrespective of the presence of transverse reinforcement (on average14%) (see Table 4).For beams without vertical shear reinforcement, loads of about twice the diagonal crack load can be sustained before ultimate failure (Vc,exp/Vu,exp of Table 4), on the average.It is also expected that the ultimate shear strength resistance for PKS is higher than Normal Concrete.Nonetheless, the contribution of transverse reinforcement to shear resistance may be slightly high in NC deep beams than PKSC deep beams at a particular level.
Comparison of the shear capacities of the PKSC and NWC deep beams can be better achieved by normalizing the ultimate shear load by its corresponding 28-day cube compressive strength of the concrete (vnorm).The results indicates that PKSC deep beams exhibit higher normalized shear strength characteristics compared to the NWC deep beams (see Table 4).This confirms the result of Alengaram et al. [8] which reported higher shear strength for PKSC slender beams compared to the control NC beams.

Shear Strength Models of Deep Beams
The traditional sectional design approach which applies the Bernoulli's assumptions of linear strain variation of a beam section and also assumes a parallel chord truss model is not applicable to deep beams.This is because, as a result of the small a/d ratios in a deep beam (a/d<2.5),a large portion of the loads is transferred directly to the supports via compression struts forming a tied-arch.Shear design of deep beams employed by several codes of practice may be classified into two major categories; those that use empirical formulae and those based on the analytical strut-and-tie models.Nonetheless, Mihaylov et al. [31] recently proposed a kinematic model for estimating the shear strength of deep beams by quantifying the individual contributions from aggregate interlock, dowel actions and stirrups.The shear transfer capabilities of a typical empirical formulae code (ACI 318-99 [32]), those of 2 strut-and-tie models (ACI 318-05 [33] and EC 2 [34]), and a kinematic model of Mihaylov et al. [31] are compared with experimental data obtained for PKSC and NC beams.These 4 codes of practice have been chosen as samples representing the two categories, as well as recently developed approach, in order to assess the validity of both categories in estimating the shear capacity of PKSC deep beams with and without vertical reinforcement.The main features of shear provisions of deep beams in the four design codes are presented below.
(w s ) E = 2.25w t cosθ + [(l p ) E + (l p ) p ]sinθ 2 (10) Where (l p ) P and (l p ) E are the widths of loading and end support plates respectively.
The load capacity of a simply supported deep beam is Where v e is the effectiveness factor of concrete taking as 0.75.
The minimum amount of shear reinforcement required in bottle-shaped struts which is recommended to be placed in two orthogonal directions is suggested by ACI 318-05 as follows Where A si and s i are the total area and spacing of the ith layer of reinforcement crossing a strut respectively and γ i is the angle between the ith layer of reinforcement and a strut.The value of the effectiveness factor drops to 0.6 if the minimum shear reinforcement defined in equation ( 12) above is not provided.

EC2
The equations for EC2 are similar to those of ACI 318-05.The effectiveness factor of concrete, dependent only on concrete compressive strength and does not consider the effect of shear reinforcement and transverse tensile strain, is stipulated as follows The top node depth is given by According to EC 2 the shear resistance due to vertical shear reinforcements is given by the lesser of V Rd,max = α cw b w zv e f cd (cotθ + tanθ) ⁄ Where A sw is the cross-sectional area of the shear reinforcement, s is the spacing of the stirrups, f ywd is the design yield strength of the shear reinforcement, α cw is a coefficient taking account of the state of the stress in the compression chord (taken as 1), z is given by 0.9d.

Kinematic Model
For the calculation of the shear strength of deep beams using the kinematic model V KM , Mihaylov et al. (2013) used the following equation: Where V ci , V CLZ , V s and V d are the shear forces resisted by aggregate interlock, by the critical loading zone, by stirrups, and by dowel action, respectively.
The aggregate interlock shear component V ci is expressed as follows: Where a ge is the effective aggregate size which equals the coarse aggregate maximum size a g for concrete strengths less than 60 MPa and zero for strengths larger than 70 MPa, with a linear transition for intermediate strengths.The crack width w is calculated according to the following: Where α o is the angle of the critical crack to the longitudinal axis of the beam at shear failure and it can be evaluated as: The vertical displacement ∆ c of the critical loading zone is calculated using the following equation: Where l be is the effective width of the loading plate and should not be taken less than three times the maximum size of coarse aggregate, a g ; α is the angle of line extending from inner edge of support plate to far edge of tributary area of loading plate.The term s max is the spacing of the radial cracks at the bottom of the section which can be calculated as follows: Where d b is the bar diameter, h is the beam total depth, and ρ 1 is the ratio of bottom longitudinal reinforcement.
The shear strength of the critical loading zone V CLZ is calculated from: Where k 1 is a crack shape coefficient which can be taken as 1 for beams with cot α ≤ 2.5 and zero for beams with cot α ≥ 2.5, with a linear transition for intermediate values of cot α.The average compressive stress f avg in the critical loading zone can be calculated as follows: The shear component resisted by the stirrups V s is expressed as follows: V s ρ w b(dcotα − l 0 − 1.5l be )f w ≥ 0 (25) Where ρ w is the ratio of stirrup reinforcement, f w is the stress in the stirrups.Quantity l o is the length of the heavily cracked zone at the bottom of the critical crack and is determined as follows: The stress in the stirrups is calculates as: Where f yw is the yield strength of the stirrups and the stirrup strain ε w is calculated as: The shear component resisted by the dowel action of the bottom reinforcement is calculated from: Where n b is the number of bars, f ye is the effective yield strength of the bars and can be taken as the yield strength of the bars f y and not be taken more than 500 MPa.At shear failure, the dowel length l k can be taken as l o given in equation 26.Results indicates that the compressive strength, fcu, and the ratio of the shear span to the overall height (a/h) as significant contributors for bias-correction, considering the selected shear strength models (Table 6).More so, ACI-318-99 produced the lowest absolute error (ABS) after bias-correction, with the kinematic model producing the least value of the root mean-squared error (RMSE), also after bias-correction.The calibrated shear models are assessed on the tested beams and presented in Table 7. Generally, the means of the ratio of the shear experimental shear failure load (Vu,exp) to the calibrated shear capacity (Vcal) were 1.0 for all the shear models except ACI 318-99 (1.02).Hence, employing the calibrated shear model of ACI 318-99 can yield conservative estimates compared to the others, particularly for design purposes.In order to better understand and determine which of the models is appropriate for design, Figure 5 shows the predicted shear against the experimental observed responses for the four shear strength models.The dotted line is used to demarcate the boundary for which the predicted response may be either above or below the experimental responses.Observations above the dotted line produces conservative estimates (predicted shear strength lesser than experimental response).A critical assessment reveals that only 3, 6, 6 and 5 of the beams had calibrated responses greater than the experimental (un-conservative) for the ACI 318-99, ACI-318-05, EC2 and Kinematic Model respectively.This suggests that the developed calibrated model of ACI-318-99 can be fairly used to quantify the shear strength of PKS deep beams during structural design.Nevertheless, it yields predictions with the lowest scatter (coefficient of variation) among the four calibrated models presented.

Conclusion
The shear behavior of reinforced palm kernel shell concrete deep beams is investigated.12 test specimen were subjected to a three-point bending test to ultimate failure.The concrete type (Palm Kernel Shell (PKS) Concrete and Normal Concrete), the shear-span to depth ratio (a/d) and the vertical reinforcement ratio (ρv) were the major variables investigated.Four shear strength models for deep beams as documented in ACI 318-99, ACI-318-05, EC2 and a recently proposed kinematic model were used to assess their predictability in relation to PKS deep beams.The major findings are:  The ultimate shear strength of PKSC beams increased with a decrease in the shear span-to-depth ratio.More so, post diagonal cracking shear resistance is greater in PKSC deep beams than beams of normal concrete.The expected improvement in post-diagonal cracking resistance of such beams averages to about 14%.
 Design shear provisions documented in ACI-318-99, ACI-318-05, EC 2, as well as the Kinematic model of Mihaylov et al. (2013) yielded un-conservative predictions for the investigated deep beams.Among these basic shear strength model, ACI-318-99 performed better than the other models.This necessitates the development of a calibration procedure to better assess the performance of these models at predicting the shear strength PKSC beams.Calibrated shear strength models revealed the compressive strength and the ratio of the shear span-to-total depth as significant influential parameters for correcting the inherent bias in the original deterministic shear strength models.The calibrated functional model of ACI-318-99 may produce conservative predictions, given this limited number of test specimens.Therefore future studies should investigate the reliability of the calibrated models, and quantifying the uncertainties in the estimated coefficients of parameters, using a much larger representative dataset.

Figure 5 .
Figure 5. Assessment of Calibrated Shear Strength Models