IELTS Academic Reading Practice Test 144

For most of the nineteenth century, the machine that converted heat into movement was the reciprocating steam engine. Steam pushed a piston back and forth inside a cylinder, and a crank turned that motion into rotation. The design was reliable, but it had obvious drawbacks. Even the best of them ran at only a few hundred revolutions per minute, and they were heavy and shook violently. That mattered because the new electrical industry needed generators spinning far faster than any piston engine could drive them directly.

The notion of obtaining power from steam by making it turn something, rather than push, is very old. In the first century AD, Hero of Alexandria described a device now known as the aeolipile. It consisted of a hollow sphere mounted on a hollow axle, with two bent nozzles on opposite sides. Steam from a boiler passed through the axle into the sphere and escaped through the nozzles, and the reaction made the sphere spin. Hero's device demonstrated a principle but produced no useful power, and it was almost certainly never put to practical use.

Practical proposals appeared much later. In 1629 an Italian engineer, Giovanni Branca, published a design in which a jet of steam was aimed at the blades of a wheel, rather like water striking a mill wheel. There is no evidence that it was built successfully, and with the low-pressure steam and rough machining of the age it could hardly have been. Only when boilers producing high-pressure steam became common in the nineteenth century did the idea become practical.

The person usually credited with turning the idea into an industrial machine is Charles Parsons. Born in 1854, the youngest son of an Irish earl, he studied mathematics at Cambridge and then served an apprenticeship in Newcastle upon Tyne. In 1884 he joined a Tyneside firm, Clarke, Chapman and Company, to take charge of its electrical department. There he was immediately confronted with the problem of driving dynamos at the speeds they required.

Parsons's answer was to divide the work. Instead of expanding steam in a single stage, he passed it through a long series of rings of fixed blades and moving blades. Each stage extracted a modest amount of energy, and the steam grew cooler and larger in volume as it travelled along. Because the pressure fell gradually, no single rotor had to withstand an extreme force or spin at an impossible rate. Parsons called the result a compound steam turbine; the principle, now known as staging, underlies almost every turbine built since.

His first machine, patented in 1884, was coupled directly to a dynamo and produced about 7.5 kilowatts at 18,000 revolutions per minute. That was perhaps thirty times the speed of a contemporary piston engine, and it made the turbine the natural partner of electricity generation. Parsons left Clarke, Chapman in 1889 to found his own company at Heaton, on the edge of Newcastle, which built turbo-generators for power stations and turbines for ships.

In Sweden, Gustaf de Laval produced a turbine of an entirely different kind in the late 1880s. Steam was expanded in a carefully shaped nozzle to a very high velocity and then directed at a single wheel of blades, an approach known as an impulse turbine. It was simpler and cheaper than Parsons's design but ran at such extraordinary speed that it had to be connected to machinery through reduction gearing. Its first widespread use was in machines that separated cream from milk.

In the United States, Charles Curtis patented a variation in 1896, in which several rows of moving blades extracted the steam's energy within a single stage. In France and Switzerland, engineers developed multi-stage impulse turbines of their own. By the early years of the twentieth century the steam turbine had displaced the piston engine in the two applications where speed and size mattered most.

The first was electricity supply. A single turbine running at high speed could drive a generator far larger than any that a reciprocating engine could turn, and because turbines were compact they fitted into city-centre power stations. Larger generators produced cheaper units of electricity, and cheap electricity in turn made electric light, trams and eventually domestic appliances affordable. As boilers, materials and blade design improved, the thermal efficiency of the best steam plants rose steadily.

The second was the propulsion of ships. Parsons built an experimental vessel, Turbinia, to demonstrate what a turbine could do. Launched in 1894 and some thirty metres long, she was fitted with Parsons turbines driving three shafts. At the naval review held at Spithead in 1897 to mark Queen Victoria's Diamond Jubilee, Turbinia threaded her way between the lines of warships at about 34 knots. She was faster than anything else afloat. The world's navies soon ordered turbine-driven vessels, and the Cunard liners Lusitania and Mauretania, completed in 1907, used Parsons turbines.

Steam turbines remain the workhorses of large-scale electricity production. Coal, gas and nuclear plants all use them, and the same aerodynamic principles are applied in the compressors and turbines of jet engines. The machine that Parsons built in 1884 has been enlarged from a few kilowatts to well over a thousand megawatts. Yet the sequence of fixed and moving blades inside it would still be recognisable to him.

Every photograph posted on a social network, every search and every card payment passes through a data centre. The term describes a building, or a floor within one, filled with rows of steel racks holding servers, storage devices and networking equipment, together with the power and cooling systems that keep them running. Few people ever enter such a place, and fewer still could point to one on a map. Yet the data centre has become the industrial plant of the digital economy, and its appetite for electricity has begun to attract the attention of governments, utilities and engineers.

Forecasts of runaway consumption have an awkward history. In the late 2000s some analysts predicted that the electricity used by data centres would soon double; in fact it grew far more slowly than expected over the following decade. The International Energy Agency has estimated that data centres alone account for roughly one to one and a half per cent of the world's electricity - about as much as a mid-sized country uses in a year. Much of the growth in demand was cancelled out by improvements in efficiency. Whether efficiency can go on keeping pace with the growing demand for computing is now the central question for the industry.

The physics has not changed. Almost every watt delivered to a processor is eventually released as heat, so the electricity bill of a facility divides into two parts: the power used by the computing equipment itself, and the power used to keep that equipment cool and to deliver current to it. Engineers express the balance as power usage effectiveness, or PUE, calculated by dividing the total power consumed by a facility by the power consumed by the computing equipment inside it. A value of 1.0 would mean that every watt reached a server and none was wasted on the way.

Cooling is usually the largest single item in that overhead. Traditionally, cold air is blown up through perforated floor tiles and drawn through the racks from front to back, so that racks are arranged in alternating 'cold' and 'hot' aisles to stop the two streams mixing. In cooler climates, operators can simply open the doors and use outside air, a technique known as free cooling; this is one reason why so many new facilities have been built in Ireland, the Nordic countries and northern China. Evaporative systems, which cool air by spraying water into it, need less electricity but consume large quantities of water - a serious drawback in dry regions.

Measurement has itself driven improvement. Because PUE made waste visible, designers began to treat it as a competitive score, and the companies that build at enormous scale have pushed their newest designs from values around 2.0 in the early 2000s to about 1.1. These hyperscale operators, whose facilities hold tens of thousands of servers, have achieved the change by sealing buildings, removing unnecessary conversions between alternating and direct current, raising the temperature at which equipment is allowed to operate, and designing their own servers and power supplies.

There are limits. PUE cannot fall below 1.0, and each further improvement is harder and more expensive than the last. Even a perfect cooling system would leave the fundamental cost of computation itself. That cost has become much more visible in the past few years, because the graphics processors used to train artificial intelligence models draw many times the power of a conventional server, and a single rack of them can demand as much as a small office block. Utilities in some regions now report that data centres are competing with housing and manufacturing for connections to the grid.

Operators have responded in several ways. The largest have signed long-term contracts to buy renewable electricity, though critics point out that such agreements do not necessarily bring new generating capacity onto the grid. A more physical answer is to stop throwing the heat away. In a district-heating scheme, waste heat from servers can be captured and piped to nearby homes and offices; systems of this kind have operated in Stockholm and Helsinki, and others have been proposed elsewhere. Batteries and fuel cells are also being used in place of the diesel generators that facilities once kept for emergencies.

Research is increasingly directed at the silicon. Moving data between chips with light rather than with electricity, an approach known as silicon photonics, would cut both the energy and the heat involved in shifting large quantities of information. More radical designs place whole racks in sealed tanks of non-conducting fluid, so that components are cooled by direct contact rather than by air. And because electricity is expensive to transport and cheap to generate in some remote places, some companies are prepared to build where the wind blows and the air is cold, however far that is from their customers.

None of this guarantees that demand will stop growing. Streaming, cloud storage and now artificial intelligence all require computation, and somebody must pay for it in electricity. But the record of the past fifteen years suggests that the sector responds to measurement and to price. The question is whether the incentive to be efficient can keep pace with the demand to be connected.

A decade and a half ago, fitness was something people did; now it is also something they record. Wristbands and phones log steps, distance, heart rate and sleep, and the software that accompanies them turns those numbers into rings to close, badges to win and streaks to preserve. The appeal is easy to understand. Exercise is a habit that people consistently intend to acquire and consistently fail to keep, and an application that nudges, reminds and rewards looks like exactly the tool the problem requires. Yet whether such products genuinely change behaviour, rather than merely documenting it, is far from settled.

Some features certainly work. Self-monitoring is one of the oldest findings in the study of behaviour change: people who keep a record of what they do tend to move closer to their goals, and a phone that counts steps automatically removes the effort that paper diaries demanded. Goal-setting, feedback and prompts have similar support. Trials of app-based programmes have generally found modest increases in walking and in minutes of moderate activity during the first weeks, particularly among people who were previously inactive.

The difficulty lies in what happens next. Reviews of this literature keep reporting the same pattern: the effect is largest early on and shrinks as time passes, and a large share of users abandon an application within weeks or months. Novelty is part of the explanation, since a new device is interesting for a fortnight and then is not. But there is a deeper problem, and it concerns the kind of motivation that the software appeals to. Most apps hand out points, badges and public rankings, and these are what psychologists call extrinsic rewards: prizes that come from outside the activity itself.

Self-determination theory, developed by Edward Deci and Richard Ryan, holds that motivation is not a single quantity but a family of different kinds. People persist at an activity when it satisfies three needs: a sense of competence, a degree of autonomy and a connection with others. Rewards can support those needs - a badge may confirm that someone is improving - but they can also undermine them. If the only reason a person has for walking is to satisfy the software, then the moment the software stops being interesting, so does the walking. Experiments going back to the 1970s have repeatedly shown that paying people for something they already enjoy can reduce their willingness to do it for its own sake.

The design of many apps pushes hard towards external control. Streaks punish a missed day, leaderboards compare beginners with experienced runners, and notifications arrive at times that suit the software rather than the user. One result can be that exercise is chosen to satisfy the number rather than the body: a short walk late at night to keep a streak alive, when rest would have been better. Clinicians have also begun to describe patients who are anxious about the data their devices collect, and who sleep badly as a result of monitoring their sleep.

Accuracy matters here too, because a measurement that people trust and that is wrong can do harm. Step counts from consumer wearables are generally reasonable, and heart-rate sensors have improved markedly, but estimates of energy expenditure - the calories that so many apps display so prominently - remain unreliable. A device that overstates the calories burned on a walk may encourage its owner to eat accordingly, while one that undercounts steps may make a genuinely active person feel that the effort was wasted.

There is also a question of who is excluded. Applications assume a smartphone, a reliable connection, a safe place to walk and a body that can be measured in the usual ways. People with disabilities, older users and those on low incomes are poorly served, and the data they generate may be sold or used to set insurance premiums. An enthusiasm for personal measurement can quietly turn a public health problem into a matter of individual willpower, which is exactly where governments have often preferred to leave it.

None of this means that the technology is worthless. When apps are designed around autonomy and competence rather than competition, and when they are attached to programmes that include human contact - a coach, a class, a group of friends - the evidence for benefit is more encouraging. The most useful features are often the least glamorous: a simple reminder, a clear record of progress, a message from a real person. What the successful examples have in common is that the software supports a decision the user has already made, rather than trying to manufacture that decision on the user's behalf.

My conclusion is therefore a cautious one. Fitness apps are a reasonable aid to people who are already motivated and who want a convenient record of what they do, and they may help some of the undecided through the first difficult weeks. But they are not a solution to inactivity, and the excitement they generate should not be mistaken for evidence that the problem is being solved. Motivation is mostly social and environmental: it comes from having somewhere pleasant to walk, from having someone to walk with, and from a life that leaves time for both.