Water resource development

Appraising the lifecycle costs of SA’s INTERBASIN WATER TRANSFER PROJECTS Drinie J van Rensburg Drinie J van

South Africa has one of the most IBTs is the Lesotho Highlands Water INCREMENTAL APPROACH sophisticated bulk water infrastructure Project (LHWP) of which planning of Phase 2 is currently underway. ase study research and secon­ networks in the world where water is often IBTs often involve pumping water Cdary data analysis was employed pumped hundreds of kilometres from areas to overcome differences in eleva- to examine the accuracy of the of surplus to areas of need through inter- tion. The associated energy costs appraisal approach, called the Incre- typically form a significant part of mental Approach, originally followed basin transfer schemes (IBTs). Research the lifecycle costs of IBTs, and it is during the planning of the Usutu- undertaken by Dr Peter van Niekerk, expected that such energy costs will Vaal Government Scheme (GWS) formerly of the Department of Water proportionally increase in future as (Second Phase) as a case study. This water has to be sourced from more IBT scheme was originally planned Affairs (DWA), has shown that the way in distant basins. It is therefore impor- and built in the early 1980s to supple- which lifecycle costs have been historically tant that a robust appraisal method- ment the water resources of the Vaal evaluated, especially in South Africa, has ology be followed when assessing the River system. The scheme shown costs of water transfers of future IBT in Figure 4 (along with other IBTs not been done optimally, which could have projects. similarly connected) consists of the severe repercussions for operational cost A conceptual inaccuracy has, Heyshope Dam on the Assegai River, estimation. for decades, been part of the water a tributary of the Usutu River, and engineering standards of practice so transfer infrastructure to convey that certain calculations are gener- water against an elevation differ- rom the 1960s South Africa ally calculated incorrectly. Over ence of 183 m over the continental has seen an increasing num- time this error has worked its way divide into the catchment, ber of IBTs being constructed into the accepted methodology of upstream of the Grootdraai Dam in Fto meet its growing water demands. evaluation of the lifecycle costs of the Vaal River. One of the most sophisticated of IBTs. This has caused a bias in the The Incremental Approach, these is the complex interconnected estimations of variable costs, such depicted in general terms in Vaal River system, which includes a as pumping costs, and made capital Figure 1, is a deterministic plan- number of IBTs constructed to serve intensive gravity schemes seem, in ning approach where the growth in the economic heartland of the coun- comparison, more attractive than demand is such that a new scheme, try. The most well-known of these they really are. an IBT in this case, is required to

28 The Water Wheel November/December 2013 Water resource development

meet requirements beyond time T1. The erratic nature observed is, if The Incremental Approach assumes anything, reinforced by the transfers Appraising the lifecycle costs of SA’s that all requirements beyond the from the Slang River. The combined capability of the existing system, i.e. average transfer of 42,5 million m3/a the shaded area in Figure 1, must reduced the discrepancy somewhat, INTERBASIN WATER TRANSFER PROJECTS be met from the new resource. The but it remains significantly less than annual water transfers therefore the average of 73,5-million m3/a will exhibit gradual growth, from water transfer originally envisaged time T1 until time T2 and then be for the period. capped by the yield capability of the Another scheme, the Vaal River new scheme. As pumping costs are Eastern Subsystem Augmentation directly related to quantities trans- Project (VRESAP), links the Groot- ferred, these annual costs exhibit the draai Dam subsystem to the Vaal same pattern. Dam. The VRESAP was completed in 2009 and made no significant con- ACTUAL TRANSFERS OF tribution during the period under THE USUTU-VAAL GWS observation. An examination of changes in he actual annual volumes of demand revealed that growth had Twater historically transferred been slower than predicted. This from Heyshope Dam are shown in would explain the lower volume in Figure 2, and compared to the origi- water actually transferred. However, nal projection, undertaken in 1981. the historic firm yield of Grootdraai Whereas the original projection was Dam had been reassessed in 2001 for a smoothly growing transfer, and found to be only 77% of the capped by the capacity of the transfer earlier estimate. The incremental scheme, the actual transfers exhib- water requirement, following the ited an erratic pattern. It was also approach as depicted in Figure 1, found that the average transfer was would in retrospect have averaged 3 26.5 million m3/a over the period 61,4-million m /a for the equivalent – only 35% of what had been envis- period, which still meant that the aged at the planning stage. actual transfers fell significantly Plausible explanations for the short – by some 30%. differences between predicted and As regards the possibility of sup- actual transfers were sought by ply constraints causing lower trans- examining possible impacts from fers, records showed that both the other resource developments in the Heyshope and Zaaihoek Dam had basin, changes in demand and source been relatively full over the period, constraints. indicating that there had been no A resource development that impediment, from a source perspec- tive, in transferring water to the Vaal did have an effect was another IBT order to set the operational regime Figure 1, top: River basin. completed in 1988, namely the for the following year. This so-called Determining pumping A detailed tracing of the annual costs with the Incre­ Slang River GWS. This scheme Annual Operating Analysis (AOA) operation of the Vaal River supply mental Approach of IBT (also shown in Figure 4) was built takes into account the state of the project appraisal. system, with the inclusion of the primarily to supply water from system (i.e. storage of each dam) as Grootdraai Dam subsystem, was the Zaaihoek Dam on the Slang at 1 May of each year. The analysis is Figure 2, middle: required to shed further light on the River to a new power station, but conducted soon afterwards. Usutu-Vaal GWS reasons for the observations depicted (Second Phase): Historic the facility to transfer water to the The AOA entails running DWA’s in Figures 2 and 3. water transfers from upper reaches of the Vaal River was Water Resources Planning Model Heyshope Dam against included in its design. Actual water (WRPM) to simulate the behaviour transfers predicted. transfers from the Zaaihoek Dam ANNUAL OPERATIONAL of the Integrated Vaal River System to Grootdraai Dam over the period SYSTEMS ANALYSES (IVRS), usually for 1 000 stochastic Figure 3, bottom: Combined historic water 1990 to 2010 were combined with time-series of flows of 20-year dura- transfers from the the transfers from Heyshope Dam to nnually, at the end of the rainy tion. This is repeated for various sce- Heyshope Dam and the obtain a more complete perspective Aseason, the DWA undertakes an narios to test the impact of variations Zaaihoek Dam on the as depicted in Figure 3. analysis of the Vaal River system in in operational rules, different growth Slang River.

The Water Wheel November/December 2013 29 Water resource development

Figure 4: Locality prevented continuous transfer at of Usutu-Vaal GWS maximum capacity during these (Heyshope Dam) within years. The policy at first was to the upper Vaal River system. keep all the water transferred in Grootdraai Dam, i.e. not to sup- port the rest of the system. When Grootdraai Dam filled completely (largely as a result of the transfers) it was decided to continue pump- ing as there were delays in bring- ing the LHWP on stream, keeping Grootdraai Dam full and the spills to augment . All transfers were stopped in December 1995 due to the system having recovered from the drought. After the drought years of 1992/93 to 1994/95 the operating projections, and possible changes to and medium term and the levels of rule for transfers to Grootdraai the configuration of the system. assurance of water supply. Dam followed a fairly consistent The WRPM is a South Afri- It was found that the state of the pattern. The two transfer projects can developed system analysis system, in particular the level of were treated in tandem, and similar simulation tool to determine the water in storage in Grootdraai Dam, recommendations made for trans- probabilistic yield characteristics of had been the dominant determinant fers from the Heyshope Dam and complex water resource systems. It in the decisions related to the trans- the Zaaihoek Dam for the year that can accommodate multiple demand fers from Heyshope and Zaaihoek lay ahead. These recommendations centres (i.e. yield channels), take dams. The state of the system at the were based on the likelihood that into account growth in demand and decision date is influenced by the water shortages may develop in the changes in land use, allow expansion inflows the dams received in the short to medium term. Basically, of the system at future dates, deal preceding period and the abstrac- interbasin transfers were made from with quality constraints, and impose tions that were made. Of these, the Zaaihoek Dam and Heyshope Dam curtailments. Its great advantage inflows are by far the dominant to Grootdraai Dam every time the is that it explicitly provides infor- factor; abstractions were more stable latter reached a certain percentage of mation on future risks associated and predictable. its full supply storage. with postulated management and During the 22-year period the For 11 out of the 16 years since intervention scenarios. It is used system storage generally remained the drought, this level was set at for operational as well as planning fairly high, except for a period 75%. During these years the starting purposes. between 1992/93 and 1995/96 when storages of Grootdraai Dam averaged Dr Van Niekerk studied annual a drought occurred. This drought 96.1%. For four of the years a 90% reports, covering a period of 22 was of such a severity that restric- rule was adopted. The average start- years, starting in 1990/91 and ending tions had to be imposed on all water ing storage in these cases was 95.2%, in 2011/12, to see how the IBTs from users on 1 April, 1995. Irrigation which was not significantly different Heyshope Dam and Zaaihoek Dam water users were restricted by 40%, to the average for the 75% rule. It were operated in the light of the con- municipal users by 20%, mines and is noticed that three of these years ditions in the rest of the IVRS during the SASOL oil-from-coal facility by occurred in the five years before the that period, and in particular that of 10% and power stations by VRESAP pipeline started at the end the Vaal River Eastern Sub-System 5%. Following exceptionally good of 2008 to augment the system from (VRESS) to which these IBTs are rains in November, 1995 the drought the Vaal Dam. This is consistent with linked. A timeline perspective could was broken. Restrictions were lifted the behaviour that can be expected be obtained of the four variables that in January 1996. of a system that is moving closer to influenced the decisions related to For the four years of the drought, its maximum safe yield capacity, with the transfers from these IBTs; the large quantities of water were trans- a growing demand and just prior to state of the system at decision date, ferred from the Heyshope and the the system being augmented. the changes in system configura- Zaaihoek dams. Problems with In the case study of the Usutu- tion and capacity, the projections in the pumps at the Geelhoutboom Vaal GWS IBT the lower quantity demand on the system for the short pumping station at Heyshope of water actually transferred, as

30 The Water Wheel November/December 2013 Water resource development

compared to what had originally the probabilistic characteristics of severely biased with respect to vari- been predicted, as well as the erratic water transfers and, following this, to able costs and that this bias leads to nature of actual transfers as com- upgrade the estimation of the costs significantly different estimations pared to the smooth growth origi- associated with such transfers. of likely lifecycle project costs. Such nally assumed, could be explained by The Comprehensive Approach differences conceivably lead to sub- tracing the annual operations of the firstly requires the integrated analysis optimal decision-making. project over 24 years. It was shown of the source basin and the receiving These findings have relevance that the Heyshope and Zaaihoek system, in combination. A simula- not only to water scheme compari- transfers were primarily determined tion model, such as the WRPM, sons, but also to the institutional by the state of storage in the Groot- must be set to record water transfers, and financial design of water aug- draai Dam, and that the latter stor- at suitable (typically monthly) inter- mentation projects, such as sea- age was dependent on the inflow vals. The model is then employed to water desalination works as well into the dam, i.e. the run-off arising obtain simulated sequences of future as the calculation of international naturally in its catchment. By exten- water transfers of IBT schemes water royalties, such as exist cur- sion therefore, the Heyshope and under consideration. With stochastic rently between South Africa and Zaaihoek transfers were dependent hydrological data input the transfer Lesotho. on the runoffs that occurred in the data sequences thus generated will In addition, the generally used catchment of the Grootdraai Dam, exhibit characteristics similar to the unit reference value (URV) measure i.e. the receiving catchment. As these actually observed water transfers of for appraising and ranking water runoffs are inherently uncertain, existing IBT projects. resource projects in South Africa hydrological uncertainty similarly With the synthetically generated also required improvement. Dr Van characterises water transfers. water transfer data as input into the Niekerk has shown that the current In the original planning of the lifecycle cost model, the analyst is approach is conceptually flawed as Usutu-Vaal IBT, the investigation into able to obtain a representation of the it fails to distinguish between water the annual operating analysis showed probabilistic characteristics of the life- transfers and effectiveness outputs. that the assumption that all incre- cycle costs of an IBT scheme under He proposed an expanded and mental water requirements, beyond consideration. Such output would improved determination of the URV the yield capability of Grootdraai form an important input into further measure. Dam, would be required to be met economic investigations, such as • This article is based on two from transfers had been incorrect. cost-benefit analysis, to ascertain the papers originally published The Incremental Approach led to feasibility of a proposed IBT scheme. in Water SA Vol 39 No 4 an overestimation of the quantities The application of the Compre- titled ‘Hydrologic-economic of water to be transferred and, by the hensive Approach was consequently appraisal of lifecycle costs same token, operational costs. The demonstrated by means of an exam- of inter-basin water transfer correct appraisal approach would be ple using the Thukela Water Project, projects’ and ‘Unit Reference to treat water transfer as a stochastic a proposed IBT to further augment Value: Application in apprais- variable. The Incremental Approach, the Vaal River system. It was shown ing inter-basin water transfer being a deterministic approach to that the Incremental Approach is projects’. Visit: www.wrc.org.za IBT scheme appraisals, does not capture the essential characteristic of future water transfers – these being of an erratic and uncertain nature. A further four South African case studies and two international case studies (in Australia and China respectively) showed that the Incre- mental Approach is still in general use today. THE PROPOSED COMPREHENSIVE APPROACH The Vaal River system combines a number of inter-basin transfer n improved approach, called schemes to water Athe Comprehensive Approach, South Arica’s economic is proposed to explicitly address heartland.

The Water Wheel November/December 2013 31