Peak-Shift Method for Notebook Computers

A Power Management Approach for Load Leveling

Masahiko Hatori

Portable Systems, IBM Japan

Abstract

Electrical generating utilities are increasingly troubled by the growing gap between the peaks and valleys of electricity demand. This paper proposes a peak-shift method for notebook computers. This method is to enable a peak-shift on devices with built-in batteries such as notebook computers. Performing a peak-shift on such devices narrows the gap. In addition, it also helps reduce $\text{CO}_2$ emissions. IBM’s ThinkPad notebook computers as a real example are introduced. Although the size of the peak-shift on each device is small, by combining large numbers of devices, it is possible to show large effects.

Keywords— ThinkPad; peak shift; load leveling; power management;

I. INTRODUCTION

Most electrical generating utilities are faced with a paradoxical problem. On one hand, they face continuous increases in electricity demand, especially for several hours of summer afternoons. On the other hand, their power plants are usually operating at much less than their capacities. They have to maintain sufficient supply capacity that always exceeds the short-term peaks of the electric power consumption.

This is particularly serious when they must construct new plants to cover the power peaks, because it is not efficient and new power plant construction is environmentally difficult.

Moving the peak of the electric power usage to off-peak hours, the “peak-shift”, is a way to address this problem.

The peak-shift has not only economic effects but also environmental effects. The electric generating utilities have been developed peak-shift devices to reduce power consumption during peak hours. The existing peak-shift devices, such as eco-ice or eco-vendor, are utilizing thermal storage systems to reduce the peak power usage. However, for other devices, such as battery-equipped devices, there was no way to perform a peak-shift.

We develop a method for a peak-shift using a built-in battery and a power management program on notebook computers.

II. DESIGN

A. Trend of electric power demand

Fig. 1 shows the electric power demand in Japan. The electric power demand has increased continuously for the past 30 years. The electric power demand increased due to economic growth, because there is a relationship between electric power demand and GDP.

The electric power demand did not decrease even though a total energy consumption (which includes oil and gas) decreased during the first and the second oil crises in 1973 and 1979. This shows that electricity is the most basic energy and the dependence on electricity is very strong.

• Figure 1. Electric Power Demand (Total energy demand, GDP, Electric power: 1965 – 1995)

Fig. 2 shows the electric power demand in a year. It was almost flat between seasons in 1967, even though a very little peak was admitted in winter. However, in 1975, the peak power demand comes to be observed around August. After 1985, the peak power demand grows up every year, then, the peak power demand was established in 1990. The peak power demand lasts for the three month between July and September.

• Figure 2. Electric power demand (year)

The peak power demand lasts for the three hours between 13:00 and 16:00. This is caused by the operation of air conditioners responding to the high temperature. In contrast, the power demand between 22:00 to 08:00 is very low. The gap between maximum load and base load increases every year. The most desirable pattern for the peak-shift is to cut the power consumption from 13:00 to 16:00 and to use more power from 22:00 to 8:00.

Because technology that stores electricity on a large scale has not been developed yet, the electric power companies have to secure enough power generation capability to cover this peak. However, the construction of these power plants creates extra costs and lowers the efficiency of electric power generation.

Fig. 3 shows the electric power demand on a summer day. The pattern is the same now and in the past, where electric power consumption is low at night and high during the daytime. However, The difference between the peaks and valleys has increased, to 10,020 MW (mega watt) in 2001 from 4,030 MW in 1975.

• Figure 3. Electric power demand (day)

B. Peak-Shift Devices

To solve this problem, the peak-shift method was developed. This idea is that it will decrease the electric power consumption during peak hours and increase the electric power demand during off-peak hours. There are two typical peak-shift devices in the market, Eco-ice and Eco-vendor.

Eco-ice is an air-conditioning equipment with thermal storage. When Eco-ice is used for air-conditioning in summer, it makes ice in the night when there is enough electric power. During the daytime, the eco-ice system uses the low temperature of the ice without using the electric-power consuming heat-pump. Conversely, hot water is made in the winter by using the electric power at night, and the eco-ice heating system uses the heat of that hot water during the daytime. As a result, the peak-shift is achieved by saving the electric power in the daytime, and by using the electric power at night.

Eco-vendor is a vending machine that is controlled according to the time. The contents of the vending machine are cool beverages. These do not freeze at 0 degrees even if they are cooled too much, though they are usually cooled down only to about 5 degrees. Relying on this characteristic, Eco-vendor overcools the drink to about 0 degrees between 10:00 and 13:00 before the peak starts. And, Eco-vendor stops running the heat-pump from 13:00 to 16:00 of peak hours. The temperature will remain close to five degrees as it gradually goes up. Eco-vendor restarts the cooling again at 16:00 and returns to the regular temperature. As a result, the peak-shift is achieved.

C. Battery power management for peak-shift

Every electronic device which includes a built-in battery (e.g. a notebook computer) is normally running on AC power when it is at a fixed place and is connected to an AC adapter. Once disconnected, the power supply will instantly switch to the built-in battery.

The transition between the AC adapter and the battery is normally triggered by the power supplied from the AC-adapter. If the AC adapter is attached, the device is normally running in AC power mode and consuming AC power.

We developed a power management method to control the switching of the power source according to the time. This method switches the power source from the AC adapter to the battery when peak demand time starts, and switches it back to the AC adapter when peak demand is past. This method allows any electronic device that includes a built-in battery to cut the AC power consumption during the peak hours. Fig. 4 shows the pattern of AC power consumption.

• Figure 4. AC Power consumption pattern

III. IMPLEMENTATION

A. Peak-shift Implementation

Fig. 5 shows power operation of a notebook computer. There are basically three operational modes. These three modes are controlled by two switches. When using the notebook computer in the office, it operates with power from the AC adapter. In this case, the switch on the AC Adapter side is turned on and the switch on the battery side is turned off.

• Figure 5. Power operation of a notebook computer

When the notebook computer is leaving the office and going out, it operates with power from built-in battery. In this case, the switch on the AC adapter side is turned off and the switch on the battery side is turned on.

When the notebook computer returning from outside to the office and connecting with the AC adapter again, then it operates with power from the AC adapter and it also recharges the built-in battery. In this case, both switches are turned on.

These operations are fully automatic and controlled by hardware logic or the micro-controller in the notebook computer.

We implemented the peak-shift first on a notebook computer (ThinkPad R31 2656-48J/C5J). Fig. 6 shows a block diagram of the system. To perform a peak-shift, we developed a “Peak-Shift Control Program”. It is a utility program which runs on Microsoft Windows 98/2000/XP. It reads the time from the internal real-time clock and controls the switches for the AC adapter and battery.

• Figure 6. Block diagram of peak-shift system

This program also controls the recharging of the battery according to the time. The battery is discharged during the peak hours, and needs to be recharged. This program can start the recharging of the battery depend on the level of peak-shift supported by notebook computer hardware. The best setting times for the start of peak-shift, the end of peak-shift, and the start of battery recharging should be 13:00, 16:00, and 03:00, respectively.

B. Three Generations of Peak-Shift

The first peak-shift PC (ThinkPad R31 2656-48J/C5J) started recharging of the battery immediately after the peak-shift ended.

This achieved the purpose of cutting the electric power during the peak hours. However, it did not achieve the purpose of using the electric power of the off-peak hours more efficiently.

When the operation of the peak-shift is ended, both the electric power that the notebook computer normally consumed and the electric power to recharge the battery were required. This consumes more electric power until the charging ends than does the normal operation of the notebook computer. We define this as the first generation of peak-shift notebook computer.

We then developed the second peak-shift PC (ThinkPad R31 2656-1MJ/1WJ/1HJ/1PJ), which improved the recharging of the battery. We made this machine start recharging after the system is shut down. This enabled it to use the power supply after turning off the computer, which is usually after business hours, in the evening or at night. We define this as the second generation of peak-shift notebook computer. As a result, it became possible for the system to recharge the battery without increasing the maximum electric power consumption.

However, in order to achieve the greatest effects of the peak-shift, it is best to recharge in the midnight time period when power consumption is at its minimum.

We developed the third peak-shift PCs (ThinkPad R40/G40) which start the recharge at midnight. We enabled the system to wake-up and start recharging by using the internal real-time clock. As a result, this allows it to recharge the battery at any time specified, even at midnight. We define this as the third generation peak-shift notebook computer.

IV. EVALUATION

A. Peak-shift effect

We measured the AC power consumption for a group of eight ThinkPad R31s (Fig. 7). This is the second generation peak-shift notebook computer. Before running the Peak-Shift Control Program, it was about 170 W on average. When running the program, the AC power consumption dropped to almost 0 W from 13:00 to 15:00. The effect of the peak-shift was measured as 170 W on average (it is about 21 W for each ThinkPad R31).

• Figure 7. AC power consumption of eight ThinkPad R31s

The number of notebook computers shipped in 2002 was 6,084 K (in Japan). So, if we assume all of these notebook computers used a peak-shift, the peak-shift effect would be calculated as 127,000 KW (kilo watt).

B. Reduction of $\text{CO}_2$ emissions

Electric power is supplied with a mix of hydroelectric power generation, thermoelectric power generation, and nuclear power generation. It is not easy to stop or reduce the output of hydroelectric or nuclear power plants quickly. On the contrary, thermoelectric power is easy to control. Therefore, a constant amount of the electric power is always supplied by the hydroelectric and nuclear power plants, and the variable part, especially during daytime, above that constant part is supplied by thermoelectric power.

Since the thermoelectric power generation burns oil or gas, it produces a lot of $\text{CO}_2$ emissions. As a result, the electric power production during the daytime produces a lot of $\text{CO}_2$ emissions compared to the electric power production at nighttime that produces little. Because the peak-shift method in notebook computers allows avoiding the use of some electric power during the daytime and allows using it during the nighttime, it helps reduce $\text{CO}_2$ emissions.

The situation for $\text{CO}_2$ emissions as a function of time is that the 08:00-22:00 daytime production averages $0.35\text{ Kg/KWh}$ and 22:00-08:00 nighttime production averages $0.28\text{ Kg/KWh}$, and the ratio is 100:80. In other words, the $\text{CO}_2$ emissions at nighttime are reduced to 80% of the daytime level.

There is some energy loss in the peak-shift method due to the charge and discharge of the battery. The loss is about 5% for a lithium ion battery. Taking this loss into consideration, the $\text{CO}_2$ ratio of daytime to nighttime is 100:84. Therefore, it is possible to expect a possible reduction of 16% of the $\text{CO}_2$ during the peak hours. (Fig. 8.)

• Figure 8. $\text{CO}_2$ reduction effect (-16%)

V. CONCLUSION

In our work a new method of peak-shift for notebook computers was developed. The reduction effects of AC power consumption and $\text{CO}_2$ emissions by this method were also evaluated.

By using this method, IBM launched the first peak-shift notebook computer (ThinkPad R31 2656-48J/C5J) into the market in February 2002. After that, IBM continued to improve the peak-shift and provide commercial notebook computer products. Although the size of the peak-shift on each ThinkPad may be smaller than for some other peak-shift devices, by combining the large numbers of notebook computers sold each year, this method will show large effects in the future.

ACKNOWLEDGMENT

The commercial version of peak-shift program was written by Hiroshi Kato of IBM Japan. Fruitful discussions with Yuzuru Takemura and Hidehiko Yokoi, both at IBM Japan are greatly appreciated.

REFERENCES

1. TEPCO Environmental Action Report, Tokyo Electric Power Company, 2001.
2. H. Suzuki and K. Yamaji, “Energy load leveling,” Japan Society of Energy and Resources, 2001.
3. T. Matsuda and K. Nagamatsu, “Development of high efficient eco-ice using new refrigerant,” Chuden Development News, No. 90, May 2001.
4. M. Hatori, “A power management approach for load leveling,” Proceedings of EcoDesign2003, December 8-11, 2003.
5. M. Hatori and T. Mito, “A method for load leveling using peak-shift control program,” unpublished.
6. M. Hatori, “System design for peak-shift computers,” unpublished.
7. M. Hatori and S. Niimi, “Peak-cut method to follow to electric power demand,” unpublished.
8. M. Hatori, “Development of peak-shift personal computer,” Mugendai, No.111, summer 2002, pp56-60.
9. M. Hatori, “Peak-shift personal computer,” Denki Jyoho, No.542, November 2002, pp68-71.
10. C. Waldspurger and W. Weihl, “Lottery scheduling: Flexible proportional-share resource management,” Proc. of Symposium on Operating Systems Design and Implementation, Nov 1994.
11. C. Waldspurger and W. Weihl, “An Object-Oriented Framework for Modular Resource Management,” Proc. of the 5th International Workshop on Object Orientation in Operating Systems, pp. 138-143, Oct 1996.
12. Heng Zeng, Carla S. Ellis, Alvin R. Lebeck, and Amin Vahdat, “Ecosystem: Managing energy as a first class operating system resource,” ASPLOS X, Oct 2002.
13. Heng Zeng, Carla S. Ellis, Alvin R. Lebeck, and Amin Vahdat, “Currentcy: Unifying Policies for Resource Management,” Oct 2002.

Hatori, “Peak-Shift Method for Notebook Computers”, IEEE International Symposium on Electronics and the Environment, pp.117-121, May 2004. Scottsdale, AZ, USAで2004/5/10発表したもの