Laserfiche WebLink
<br />(' <br /> <br />f <br /> <br />" <br />, <br /> <br />Table 3. Typical central praceaser unit (CPU) time break~down by job <br />function. <br /> <br /> Pe rcent <br /> CPU see of t.otal <br />1. Initialize 3.6 9.5 <br />Z. Read hydrology tape 5. I 13.5 <br />3. Surplus analysis. .3 .8 <br />4. Read scratch hydrology Z.Z 5.9 <br />5. Ba.lance the iystem 17.7 46.8 <br />6. Route salt 7.6 ZOo Z <br />7. W rite to tape edit <br />8. Print summaries .1 .3 <br />9. Read transaction cards .9 Z.3 <br />10. Print shortage messages .3 .7 <br /> Total time used 37.8 100.0 <br /> <br />'I~ble 4. Typical model run costs for Z6-year run. <br /> <br />Output Option <br /> <br />Cost <br /> <br />(a) Annual values of flow and salinity $12 <br /> <br />(b) Option (a) plus writing all information for the <br />tape edit analysis $17 <br /> <br />(e) Write all output information plus tape edit file $50 <br /> <br />(NOTE: Writing all output is very costly - this. <br />output is normally used only for debug purposes.) <br /> <br />The cost for generating a typical 30-year trace for the present node <br />and sequence point setup is less than $4. The cost for a typical tape <br />edit program run is about $10. <br /> <br />162 <br /> <br />f--' <br /> <br />,- <br /> <br />tOP'" <br />. ~.. <br /> <br />Development Problems <br /> <br />A number of problems have arisen as the simulation model and <br />associated programs were developed. The most significant of these <br />are listed here hopefully to provide guidance to others who are' develop- \ <br />ing large complex computer programs and simulation models. <br />Adequate program documentation was not accomplished during <br />model development. With the turnover of personnel associated with <br />this modeling effort, continuity of work was thus made quit.e difficult. <br />It is recommended that for future computer program development ot <br />this type that documentation be carefully written as a first step and the <br />program then developed to accomplish the desired result. <br />Because of an.ope'ratio'nal application to the Western U.S. Water <br />Plan Studies both program documentation and systematic debugging <br />efforts were delayed. The operational application was an excellent <br />learning and debugging process lor that set of conditions, but did not <br />test the range needed to, adequately debug the model. <br />A number of technical problems associated with generation of syn- <br />thetic streamflows were liolved. Significant problems still remainlng <br />relate to analysis of the historic data base for statistics to be used in <br /> <br />generation, generation of salinities {or intervening inflows, and correla- <br />tion' of these intervening flows with other !lows in the basin. <br />Analysis of the historic data is complicated by development in the <br />basin which may cause different periods oi streamflow records to be <br />incomparable. Corrections must be made to bring the data to a unit,orm <br />depletion level. Also the length of available records varies greatly <br />from station to station within the basin. <br /> <br />Generation 01 salinities within a reach 01 stream are based on <br /> <br />statistics derived from net values 01 !low and net values 01 salinity. <br />These net values are made up of both historic inflows and historic out~ <br />1lows (demands) from a reach. If in1low and outflow from a reach are <br />nearly the same the net inflow to the reach is essentially zero. There <br />may be a movement of salt into or Qut of the re.ch but it it> difficult to <br /> <br />163 <br /> <br />!. <br />