Measuring approximately 385km north-south at the longest distance, and 143km east-west at the widest section, Taiwan has an interesting geographical condition. The island is home to five major mountainous ridges running in the north-south direction with mountains extending eastward and westward. The low lands between any two
adjacent mountains form scenic valleys where rivers are formed. Generally, all rivers originate from the peak of each ridge, snake through the valleys, and run across sporadic plains to reach the Pacific Ocean on the east coast and Taiwan Strait on the west coast. Because of the high ridge peaks and steep valley basins, all rivers are short and steep causing rapid flows during stormy days, particularly in a typhoon season.

The geophysical condition in Taiwan is also remarkably unusual. The island not only sits on a high frequency seismic belt at the furthest point of the west border of the Philippine tectonic plate, but also in the Pacific typhoon paths. As a result, the annual frequencies of earthquakes and typhoons are exceedingly high. Engineers who design dams in Taiwan are required to carefully consider the unfavourable geographic conditions and pay close attention to any potential abrasion-erosion problems.

Recent surveys indicate that there are at least 1385 dams and 69 reservoirs in Taiwan, with practically all suffering abrasion-erosion problems such as surface peel-offs, concrete disintegration, and rebar exposures [9].

Because of this, it became necessary to find a viable technology to control abrasion-erosion damages on these hydraulic structures. A research team, with previous experiences on high performance concrete studies, envisioned the benefit of using low cement high performance concrete (LcHPC) to form the top layer of a Sabo dam to take advantage of the low hydration heat upon using less Portland cement. A proposal to conduct full scale field tests to compare the erosive resistibility of LcHPC against that of normal concrete (NC) was submitted to the Council of Agriculture of Taiwan in 1999. The proposal was approved and a Sabo dam was constructed in year 2000. The proposed test programme was carried out within a period of 22 months and the entire project was completed in 2002. The obtained data have been carefully evaluated and the results are discussed in the following article.

Sabo dam

Sabo is a combination of two Japanese words; Sa and Bo, The first word means sand and the second means protection. Sabo dam is literally a check dam in English, implying a dam for abrasion-erosion protection. A Sabo dam is designed to protect the downstream riverbed from hydraulic scouring and undesirable erosion [10]. A Sabo dam for the proposed investigation was constructed in the year 2000 in the Nao-Liao basin in a deep mountain area of Chia-Yi County, where the annual flow rate of water borne mud and stones is severe.

The Nao-Liao basin covers 3.72×106m2 land area of which 99% is forest terrain, stretching from an altitude of 1300m above sea level at the top to 450m at the lowest point and forging an average slope of 0.53. Many up-and-down winding creeks merge to form the main body of approximately 3.5km long rivers that end at shores of the Taiwan Strait [11].

The geological condition along the river is of an obvious loam, clay composing of 50% organic substances, 37% inorganic matter like rocks, gravels, and soils, and fractions of undetermined percent of metallic substances [4]. The upstream landscape consists mainly of hillside sparsely covered by vegetation and other agricultural plants.

Why LcHPC?

In designing concrete dams and/or spillways one must consider the potential problems associated with scouring, cavitations, and abrasion attributed to a rapid flow and impacts of water borne particles during a rainy season. Aside from hydraulic solutions, the mechanical properties of concrete play a significant role in enhancing the performance of a hydraulic structure. The following points set the course for the authors’ studies:

• Construction of a Sabo dam conceivably involves a large quantity of concrete, either ready-mixed or on-site mixed. Since hydration reactions of Portland cement are exothermic, it causes the liberated heat to gradually build up in concrete while it hardens. The temperature rise within the mass depends on the rates of heat liberation and dissipation. The temperature distribution within the mass dictates the thermal gradient that causes tensile stresses to develop. If the developed tensile stress exceeds the tensile strength of concrete at any specific points, unnoticeable internal cracks develop [12]. They are small enough to be called micro-cracks.

• Ambient temperature and moisture are also influencing factors for the crack formation in hardened concrete. For example, a sudden air temperature drop may cause surface cracks particularly in the early stage of hardening [12]. Understandably low moisture air induces rapid evaporation of mix-water and results to incomplete hydration and low concrete strength.

• Drying shrinkage is a common problem associated with the amount of water used in the concrete mix and moisture of the exposed air.

The loss of water causes a volume reduction in concrete of which the non-uniform volume change in a stereo perspective triggers cracks in concrete. The crack formations are intimately related to the cement type and the mechanical characteristics of the hardened cement paste. This understanding led the research team’s attention to the benefits of using low cement together with appropriate amounts of fly ash and slag powder, commonly known as a high performance concrete. The obvious advantage of using LcHPC is the low rate of hydration heat liberation that can significantly reduce the thermal gradient, a culprit for micro-crack and crack formations. This concept motivated the research team to investigate the erosive resistibility of low cement high performance concrete.

Test programme

Experimental Sabo dam

Figure 1a shows an elevation view of the Sabo dam. Width at the top is 27m and the depth measured from the top to the river bed is 7.5m. The spillway of dam is 2m deep and 13m wide. The centreline of the opening coincides with the centreline of the dam. At the centreline of the opening, a 2x2m channel was installed to control water flow during the low precipitation season and has been used to separate the 2.5x2m and 3x2m test blocks on each side of the channel.

Test blocks

The basic strategy of this investigation is to validate the erosive resistibility of LcHPC by comparing field abrasion-erosion test data. A total of four test blocks were cast for the proposed test programme. Figure 1b shows the block layout, in a left-to-right sequence that starts with NC and followed by LcHPC, open channel, NC, and LcHPC. The LcHPC blocks were cast with LcHPC for the top 2m and the remainder with NC. The water/cement ratio of both LcHPC and NC used is the same – 0.5. Concrete for the dam body below the test blocks was designed for 13.72MPa.


The basic concrete ingredients include Type I Portland cement, river sand, and crushed basalt rocks. Type I Portland Cement in compliance with the CNS 61 requirements was used, while the river sand used has a fine modulus of 2.98, specific gravity of 2.66, and an absorption rate of 2.5%. Coarse aggregate used has a maximum size of 20mm, specific gravity of 2.64, absorption rate of 1.2%, and a dry-rodded unit weight of 1682kg/m3. Admixtures such as Class F fly ash with a specific gravity of 2.31 from Taiwan Power plant in Taichung, ground granular blast furnace slag with a specific gravity of 2.89, and super-plasticiser complying with the ASTM C494 type-G with a specific gravity of 1.1 were used in making low cement high performance concrete.

Mixture proportions

It is a common practice in Taiwan to use 20.6MPa concrete for Sabo dam construction which generally requires a cement content of 400kg/m3 or more, a maximum aggregate size of 25mm, and a water/cement ratio of 0.5. One of the themes in this investigation is to use less cement to control hydration heat in order to minimise the occurrence of micro-cracks in concrete during the hydration process. For this reason a cement content of 325kg/m3 was used together with specific amounts of fly ash and slag to make LcHPC. In this study, the LcHPC used was water/cementitious material (w/cm)=0.32) and a maximum aggregate size of 13mm were used in the mix design as shown in Table 1. The final selection of the mixture proportion for construction was based on a series of trial mixes to optimise the economical material usage and durability of concrete. The material properties of freshly mixed concrete were also measured. Table 2 shows the measured slumps and slump flows of LcHPC.


The mixture proportions as shown in Table 1 were given to the ready-mix contractor for the actual concrete production. Further tests on slump, slump flow, unit weight, and air content were carried out at the project site and the results obtained closely agreed with those measured in the laboratory.

The delivery time from the batching plant to the construction site was approximately 45 minutes. An unfavourable terrain condition required the ready-mixed truck to discharge the concrete through a 9m long chute positioned on a 40 degree slope bank into a large steel bucket, which was then transported by an excavator to the specific locations for casting as shown in Figure 2. Segregation and bleeding were observed during the entire placing process of NC but not of LcHPC. This is a strong indication that the workability of LcHPC was better than that of the companion concrete.

Series of 10cm by 20cm cored samples were taken after 28 days and 56 days. Compression tests were subsequently conducted to determine the strengths of LcHPC and normal weight concrete. The measured compressive strengths were 56MPa and 69Mpa for LcHPC and 23.5MPa and 25.9MPa for the NC.


As soon as the experimental Sabo dam was completed in 2000, an observation programme began. First, a flow rate of approximately 640m3 /sec and sedimentation of 68m3 /sec were determined. A flow rate of 640m3 /sec reflects a significant erosive dynamic energy and the destructive power of the water borne sand, stones, and rocks. Figures 3 and 4 show a bank collapse and severe concrete abrasion-erosion, respectively. In general, the NC suffered extensive crack damages, whereas LcHPC was subjected to sporadic chip-offs. This supports the theory that concrete of inferior strength exhibits a lower erosive resistibility than a high compressive strength concrete.

Measuring abrasion-erosions

Two elevation benchmarks were established for the reference use; one on the upstream and the other downstream of the Sabo dam. A grid system of 30cm in the direction perpendicular to the water flow and 10cm in the direction of water flow were marked on the surfaces of test concrete blocks. A level was used to shoot a level rod positioned at the grid nodes to record the direct readings that were then processed to yield the elevations of each grid node. The level measurements were frequently taken throughout the entire test duration but not all are used in this report.

By comparing the measured level readings with the initial reference readings, the abrasion-erosion depths of test concrete blocks were obtained. For simplicity reasons, only data taken in the 3rd, 7th, and 22nd months are reported as shown in Table 3.

It is clear that LcHPC experienced much less damage than its companion NC. Figure 5 shows LcHPC damage of about 50% that of the NC after seven months of flow test. Figure 6 reveals the erosive damages that both types of concrete blocks suffered after 22 months of flow test. Figure 7 illustrates significant chip-offs of coarse aggregates and rebars exposures for NC but insignificant damages were found on the LcHPC blocks.


The authors believe that erosive resistibility of a hydraulic structure depends on the compressive strength and density of concrete used and the distribution of micro-cracks developed during the hydration process. This study concludes that LcHPC blocks developed a greater erosive resistibility than its companion blocks made with NC. It was found that the lower cement content of LcHPC contributed to less hydration heat, which in turn minimises micro-crack formation.

LcHPC made with admixtures such as fly ash and slag developed higher density and less porosity than those of the NC. The dense constituents strengthen the binding forces between cement paste and aggregates and enhance its ability to resist abrasion-erosion attacks. That the measured abrasion-erosion depths of LcHPC blocks after seven and 22 months of flow exposure were approximately 33 to 50% and 17 to 25% that of the NC is considered very promising. After 22 months of flow tests, the NC blocks experienced severe concrete disintegration and rebar exposure, whereas the LcHPC test blocks maintained their integrity within an acceptable tolerance.

The Sabo dam investigation concludes that LcHPC behaves similar to roller compacted concrete – it generates low hydration heat to minimise micro-crack. This finding deserves attentions from engineers and researchers in the related fields. The authors are Yu-Wen Liu and Jenq-Chuan Liou, Department of Civil and Water Resources Engineering, National Chia Yi University, and Tsong Yen, Department of Civil Engineering, National Chung Hsing University.


Table 1
Table 3
Table 2