Winds of change at Eskom's Matimba plant21 January 1999
The lessons learned from operating the direct air-cooled condenser installation at Matimba – the world's largest such installation – should prove valuable to future multi-unit power stations employing this technology.
With water such a scarce commodity in most parts of South Africa, it is obviously more advantageous for the country to utilise the available water to provide for the basic needs of its population rather than to evaporate large quantities of it into the atmosphere. Eskom, the largest power utility on the African continent and South Africa's national electricity producer, has taken unprecedented steps to conserve this essential commodity.
The past two decades saw Eskom committing some 10 500 MWe of generating plant to both direct and indirect dry cooled systems, and as a result, Eskom is today a world leader in utilising dry cooling technology in large power stations. A stalwart in Eskom's stable of dry-cooling plants is Matimba, a coal-fired power station in the dry Northern province of the country. The station has six units, each capable of generating 665 MWe. A forced draught direct air-cooled condenser (ACC) serves each unit.
These condensers, supplied by GEA Air Cooled Systems, represent a major advancement in the application of dry cooling technology for power generation. At the time of design and construction, the Matimba air-cooled condenser plant was ten times larger per generating unit than any contemporary direct air-cooled condenser plant, and still remains today by far the largest of its kind in the world. The use of dry cooling has resulted in an average specific water consumption of 0.17 litre/kWh at Matimba. This represents an annual saving of some 45 million m3 of water when compared with a wet cooled power station of this size, enough to sustain a large town of about 150 000 inhabitants.
Typical ambient conditions at the power station site near the town of Ellisras are as follows: temperatures range from -4°C to 42°C; the prevailing winds are east to north-east for approximately 70 per cent of the time; wind speeds are mostly moderate (3m/s) with occasional stormy conditions. The annual rainfall averages 430mm. The condensers were designed for an average temperature of 20°C.
The plan area of the total condenser plant is 35 700 m2, measuring 510m x 70m. The heat exchanger platform and fan deck are supported by 72 concrete columns at an elevation of 45m. The structure incorporates 288 fans of 9.1 m diameter. The fans displace close to 200 tons of air per second to reject some 5400MWt of heat with all six units on load.
The station is built with the six units in an in-line configuration typical of most South African power stations. The six condensers are arranged in tandem with no gaps in between and placed adjacent to the 550 m long turbine hall. The air inlet side of the condensers faces east to take advantage of the prevailing winds.
Although the fans are at an elevation quite adequate to allow sufficient free inflow of air on the open sides, this particular overall plant layout contributes to restrict the free inflow of air to the condenser under certain wind conditions. Somewhat unexpectedly, the units at the north and south side open ends (units one and six) are actually most affected although they have two open air inlet sides per condenser.
No forced outages were experienced due to mechanical or structural problems of the condensers in the eleven years since the first unit came on line, and no major failure is envisaged within the remaining 25 to 30 years of station life.
However, as is to be expected with a first installation of such magnitude, some teething problems were initially experienced. Adverse effects of particular wind conditions manifested in the form of condenser performance losses. Some 12 turbine trips occurred during the first 40 unit-years due to turbine backpressures increasing rapidly. Although the total losses amounted to an installed capability loss factor of only around 0.68 per cent per annum, the National Grid controllers were not comfortable with a situation where they could unexpectedly be challenged with the loss of a large generating unit.
The biggest concern was therefore not energy losses, but rather the trips that occurred occasionally in summer with little or no warning under sudden turbulent atmospheric conditions. The high-risk period is usually of short duration and occurs during the windy period preceding a thunderstorm with winds gusting at speeds in excess of 20m/s.
Trips of this nature occurred when the rate of increase in turbine exhaust pressure exceeded the rate at which the unit controls could respond to adjust the load and steamflow to maintain the condenser pressure within safe limits. Manual interference often caused the turbine bypass valves to open and dump high-energy steam directly into the condenser. This aggravates the situation and a trip becomes almost inevitable, either through HP turbine exhaust overheating or high condenser pressure.
The effects of wind direction can be summarised as follows:Easterly winds enhance the airflow into and through the fans and heat exchanger. The ACC acts as a large airscoop, allowing the kinetic energy in the wind to provide a positive pressure underneath the fan deck platform and the fan intakes. As easterly winds are the prevailing winds the plant was orientated with this advantage in mind, and good condenser performance is experienced on the whole. Westerly winds create turbulence in the wake of the 120 m high boiler houses and the turbine hall. The air circumventing the station buildings, concurrently with hot air rising from the finned tubes, fosters a complex flowfield around the ACC. The fans subsequently experience a 'breathing' problem. With the local atmospheric pressure now higher above than underneath the heat exchangers, reverse flow will take place at any fan that may be out of operation, feeding hot plume air to the underside of the platform and the intakes of the fans in service. High velocity air streams form at the northern and southern air intake sides of the platform to develop into vortices quite similar to strong whirlwinds. Horizontal vortices also form behind the windwall on the downwind side of the platform.
All these effects of westerly winds reduce the flow of cooling air through the ACC and also cause part of the hot air to be drawn back down over the edge of the platform in a process known as plume recirculation.
Winds blowing steadily from a westerly direction are detrimental to ACC performance but they do not present the same high trip risks as turbulent winds do.
Tackling the problem
To enhance the condenser's performance under adverse conditions, the following options were considered: installation of a wet cooling system in parallel to the dry cooling; wet deluging of the ACC heat exchangers; enlarging the ACC capacity; and modifying the ACC air side configuration. The cost estimates of the different options ranged from some US$ 30 million for additional wet cooling, to some US$ 0.35 million for limited configuration changes.
Several tools were developed to investigate the effect of wind on ACC performance deficiencies. A site weathermast was erected to monitor the local ambient conditions and in addition an extensive temperature and velocity sensor matrix was installed underneath the ACC platform to monitor the fan inlet air conditions. The combined information is processed to produce a visual indication in the unit control room of the air conditions that affect condenser performance. The National Weather bureau also helps by providing a daily detailed forecast of site conditions expected for the next 48 hours to assist the production planning of National Control.
Plant and process parameters were also identified that made plant operation at high backpressures arduous. These included turbine cold end conditions, condensate polishing plant temperatures, turbine steam flow and the turbine gland steam system. It became clear that it was possible, with some adaptations, to operate the units at condenser pressures higher than previously regarded to be the safe limit. However a more stable condenser operation under stormy conditions was essential to minimise trip risks.
Investigations of wind effects on the ACC included site experimentation and computational fluid dynamic (CFD) simulations. Two independent institutions with CFD capabilities were tasked to determine the dynamics of the ACC flow fields under various ambient conditions and different plant configurations. Each used its own software and code to solve for fan performance, and each specified its own boundary conditions. In all, 24 different modifications to the condenser configuration were proposed and simulated.
One modification, which was viewed with a fair amount of consensus, entailed the closure of the open 'V's on the western side of the platform and removal of the cladding between the turbine hall roof and ACC platform, with the aim of increasing the breathing capacity of the ACC and simultaneously reducing recirculation. This option is referred to as the "windwall modification".
In December 1992, the windwalls of unit one were modified as an experiment. A decrease in the effect of wind transients was experienced, but was not enough to warrant operation closer to the backpressure trip limits.
The results of the two independent sets of simulations showed that deterioration in fan performance contributed more to ACC losses than recirculation of the hot air plume. They however did not agree on the benefit of the windwall modification. Using site experimentation data on pressure and flow in the turbine hall and area below the ACC, calculations indicated that removal of sections of the cladding in the openings of the turbine hall concrete wall would further improve the airflow to the ACC. Using the turbine hall effectively as an air distribution manifold, air will flow through the turbine hall to the low-pressure regions below the fan platform.
Verification runs with CFD showed that such a modification could improve the net result of the simulations in favour of the windwall modification. The turbine hall and windwalls of the ACC were subsequently modified.
To evaluate the effectiveness of the modifications, the performance of the plant as related to weather conditions for three years prior to, and two years after, the project completion was scrutinized, using an index developed to measure the sensitivity of the plant to ambient coditions.
The study shows that a step change in performance occurred during 1995, the year during which most modifications were carried out. No ambient related turbine trips have occurred during the past 12 unit-years, and a decrease of some 60 per cent in the sensitivity of Matimba to forced vacuum load losses due to adverse ambient conditions has taken place.
The improvement came about as a result of relatively modest and cost-effective changes to the plant configuration. The cost was less than 0.5 per cent of the capital plant installation cost, and reduced condenser related installed capability losses from 0.68 to 0.3 per cent per year. Important lessons were learnt concerning the effect of relatively simple plant modifications in terms of improving air-cooled condenser performance. It is believed that future multi-unit power stations employing air-cooled condensers of similar arrangement, will benefit from the experience and knowledge gained at Matimba.