OPTIMIZATION PROJECTS

Satellite Power subsystem design, Saurabh Jain

A fundamental power subsystem consists of (Figure 1)

  1. A source of power generation
  2. A medium of secondary storage
  3. A means of directing power to the storage
  4. Modules for regulating and gauging power that goes to and comes from the storage
  5. Protecting spacecraft loads from over current/ voltage
  6. A means of coordinating the above operations, storing data and communicating with C&DH

Figure 1: System block diagram

A. Sources of power generation

There are five major ways to generate power on a spacecraft: solar energy, fuel cells, batteries, nuclear, and microwave. Solar power systems can be subdivided into solar photovoltaic and solar dynamic power generation systems. Fuel cell power systems include non-regenerative fuel cells, and regenerative fuel cells that are currently under development. Battery can be both rechargeable and non-rechargeable types. Nuclear systems can be divided into small-scale power supplies (radioisotope thermoelectric generators) and large-scale power supplies (multi-megawatt reactors). The choice of an appropriate power system depends on the amount of power required, the duration of the mission, constraints on mass and volume, and the impact of the system's hardware on the spacecraft design.

Figure 2: (Courtesy http://www.tsgc.utexas.edu/archive)

Solar Energy

  1. Solar Arrays

    This article primarily focuses on power generation and distribution using solar arrays as they have been used on VIKSAT1.

    Solar arrays consist of a large number of individual solar cells arranged on a substrate which convert solar energy into electric power by photovoltaic conversion. The solar cells are made in various shapes and sizes which put out relatively low current and voltage. Solar array deployment began with the drum-type spin stabilized vehicle where 40% of the array was exposed to the sun at any one time. Deployable paddle-like arrays evolved from the need for increased power outputs. With the technical evolution of thinner solar cells, a variety of roll-out and fold-out solar arrays have been designed and demonstrated. With the advent of flexible solar arrays, a much larger array area can be packaged for the same mass of a paddle-like deployable array. Solar cells are connected in series to maximize voltage and in parallel for current.

    To minimize power losses with a single cell failure, the solar array cells are connected in a series parallel ladder network. The current-voltage (I-V) characteristics of solar cells are of importance in the design of solar arrays. From the I-V plot (Figure 3), an array can be designed for minimum mass and maximum efficiency at the maximum power point (MPP). With the need for more power, silicon (Si) as the primary semiconductor material in solar cells will need to be replaced by a higher-efficiency material. Gallium-arsenide (GaAs) is that proposed material, which is 40% more efficient than silicon (Si cell efficiency is 18.5%). GaAs is also more resistant to radiation than Si.

    Figure 3: Characteristics of solar cells

  2. Solar Dynamic Systems

    To provide higher efficiencies for solar power production, the development of space solar dynamic power systems has been proposed. The difference between solar photovoltaic and solar dynamic power is the power conversion technique. Instead of direct conversion of solar power into electricity as with solar photovoltaics, solar dynamic systems use solar power to heat a working fluid to drive a heat engine which is used to generate electricity. The advantage of solar dynamic systems over solar photovoltaic systems is that dynamic systems in general have a higher thermal efficiency and can be used for higher power levels. A solar dynamic system consists of four basic components, the collector/ concentrator, receiver, radiator, thermal storage material, and the heat engine. The power conversion cycle can be any of the common thermodynamic cycles: Rankine, Brayton, Stirling

Fuel Cell Power Systems

A fuel cell is a device that directly converts the chemical energy of reactants (a fuel and an oxidant) into low-voltage electricity, via electrochemical reactions. A fuel cell is thus similar to a conventional chemical battery. The main difference is that in the ordinary battery, the “fuel” is the built-in expendable electrode. When this electrode is depleted, the battery is either “dead” or requires recharging in order to restore the chemical state of the electrode. A fuel cell is a converter only, using an external fuel supply.

Different Types of Fuel Cells

There are three different kinds of fuel cells with possible applications in the aerospace field:

Alkaline electrolyte fuel cells: The fuel cells used so far in the space program have all used aqueous alkaline electrolytes, which have excellent electrochemical properties and behave well with hydrogen as a fuel. Aqueous alkaline electrolyte systems have low activation energy for the cell reactions. They therefore have high power output even at below-ambient temperatures. This type of fuel cell, however, is not suited for use as a truly regenerative cell (i.e. without an electrolyser) due to material problems. The alkaline fuel cell system with a separate electrolyser appears to be best for shorter missions (<5000 hours).

Solid polymer electrolyte These fuel cells are fully regenerative (i.e. can operate in both charge and discharge modes). They are less massive than RFCs using an alkaline electrolyte, but also have lower efficiency. Studies have indicated that for longer missions (>1 year), the acid solid polymer electrolyte RFC may be most appropriate.

Solid oxide electrolytes: These materials are semi-permeable ceramics that have the ability to conduct an electric current by the passage of oxygen ions through the crystal lattice at sufficiently high temperatures. The electrolyte material used is mostly Zirconia. This type of fuel cell is fully reversible and will reach high specific energies (³l kW/kg). The major problem is the design of an efficient thermal system to maintain the 1000C operating temperature.

Battery Power Systems

Batteries have been in use for spaceflight applications since the flight of Sputnik. Since that time, batteries have matured from non-rechargeable one-use power systems to rechargeable multi-use backup power systems. In the early years of spaceflight, relatively short flight times encouraged the use of batteries as a primary source of power. As the mission durations grew longer, solar and nuclear energy took as the primary sources. This did not, however, close the door on battery use. In fact, this development sparked the need for batteries as secondary power sources.

These batteries provided power when the primary source could not. For example, when a satellite with solar panels enters a period of eclipse, batteries provide power until the satellite emerges from the occultation. Even with nuclear power sources, there are circumstances under which the peak electrical load is greater than the normal operational load. Instead of designing a nuclear reactor that operates at the peak power, a battery subsystem can provide the excess power and the nuclear reactor can be downsized for normal operation.

Battery consists of several subunits called cells. Each cell is an identical unit which can be considered a "black box" with a positive terminal and a negative terminal. Within the black box, the terminals connect to electrodes which reside within a bath of electrolyte. Energy is stored in the cell using an oxidation-reduction reaction. This reaction uses the interaction between the electrodes and the electrolyte in which electrons are transferred from the oxidation reaction at one terminal to the reduction reaction at the other terminal. This stored energy can then be tapped by reversing the reaction and letting the current flow through the load. Batteries differ due to the materials used in the design. Different electrode electrolyte combinations produce different oxidation-reduction reactions, and, therefore have different cell characteristics.

B. Secondary Storage

As a means of supplying power to the spacecraft during eclipse, secondary storage batteries to flywheels can be employed. Batteries are most commonly used. VIKSAT1 used a Lithium ion battery. Advantages of using LI-Ion battery over the traditional NiCd or NiH2 are manifold. Li-Ion has a high power density, has a relatively flat discharge curve, high voltage and do not exhibit memory effect. VIKSAT1 has a 96W (12V, 8Ahr) battery. It is made up of 12 cells. 3 cells are connected in series to form a string and there are 4 such strings connected in parallel. Each cell has a capacity of 2Ahr. The pack has been configured in such a way that the nominal depth of discharge (DOD) in most of the discharge cycles is less than 40%. Usually the cells are secured in a battery box. The battery box has provisions for draining any leaking electrolytes, taking temperature and pressure measurements and containing any fire/spark. It is a modular and a quickly replaceable assembly.

Figure 4:Battery charging characteristics

C. Charger

Performance of Li-Ion batteries comes with a cost. Charging them requires constant monitoring and regulation. To achieve maximum performance, the cells constituting the battery must be balanced. The “constant voltage/constant current” method is used to charge lithium ion batteries.

Figure 5: Charger configuration

Some key points to remember about these batteries are:

1. Charging the Batteries

  • Charge Voltage
    The maximum voltage is 4.2 V x the number of cells connected in series.
  • Charge Current
    We recommend 0.7 C. (C = Rated capacity)
    When the voltage per cell is 2.9V or less, charge using a charge current of 0.1CmA or less (Contact manufacturer for information regarding pulse charging.)
  • Charge Temperature
    The batteries should be charged at temperatures between 0°C and 45°C.

2. Discharging the Batteries

  • Discharge Current
    The current should be maintained at 1.0 C or less
  • Discharge Temperature
    The batteries should be discharged at a temperature between -20°C and +60°C.
  • Discharge Termination Voltage
    Avoid discharging at voltages less than 3.0 V per cell. Over discharge can damage the performance of the battery. Equip the unit with a mechanism to prevent over discharge.

Charging on VIKSAT1 is done using a switch mode charger.  The primary advantage of a switching charger over a linear is the heat dissipation which is of major concern in spacecrafts. Figure 5 shows the charger configuration used.

D. Monitoring

Single lithium-based cells require monitoring so that cell voltage does not exceed predefined limits of the chemistry. Series connected lithium cells pose a more complex problem: each cell in the string must be monitored and controlled. Even though the pack voltage may appear to be within acceptable limits, one cell of the series string may be experiencing damaging voltage due to cell-to-cell imbalances. We have used charge shunting to balance the cells in VIKSAT1.

Figure 6: Cell balancing and monitoring setup

The charge-shunting cell balancing method selectively shunts the charging current around each cell as they become fully charged (Figure 6). The shunt resistor R is sized to shunt exactly the charging current I when the fully charged cell voltage V is reached. If the charging current decreases, resistor R will discharge the shunted cell. Large power dissipation is avoided as we use a stepped current charger. Since VIKSAT1 has a battery with 12 cells, it is equipped with 12 shunt resistors and FET switches. All the switches are associated with drive electronics and isolators to be able to switch the FETs at a reference other then ground. Every string of 3 cells has a set of discharge and charge FETs that can be used to pull the string out of operation in case of overcharge, over discharge or over current. Every string has a current sensor for monitoring current.

A fuel gauge is used with the battery for estimating the available charger. Although lithium-ion is memory-free in terms of performance deterioration, batteries with fuel gauge exhibit "digital memory". Here is the reason: Short discharges with subsequent recharges do not provide the periodic calibration needed to synchronize the fuel gauge with the battery's state-of-charge. A deliberate full discharge and recharge every 30 charges corrects this problem. Letting the battery run down to the cut-off point in the equipment will do this.The simplest fuel gauge will involve integrating the charge flowing in or out of the battery, there by keeping track of the remaining charge. But this does not account for the self discharge or temperature variations. To overcome these uncertainties VIKSAT1 employs a combination of fuel gauging techniques and averaging them for the final result. Compensations for battery temperature, self-discharge, and rate of discharge are applied to the charge counter to provide available capacity across a wide range of operating conditions. Battery capacity is automatically recalibrated, or learned, in the course of a discharge cycle from full to empty.

E. Load Protection

It is necessary to protect the spacecraft loads from over voltage or over current. VIKSAT1 uses separate circuit breakers for each load. Circuit breakers on VIKSAT1 have the capability of auto resetting after a fault condition. They can also be controlled from the power subsystem main controller. After a trip the breaker goes into an alert condition, in that it is allowed to reset for three consecutive trips in a stipulated interval after which, if the 4th trip occurs in that period, the power to the subsystem is killed. If in the stipulated interval, less then 4 trips occur, the trip count is reset and normal operation resumes. If the power is killed to a subsystem, it can be resumed only after the C&DH commands it. This overcomes the problem of drawing C&DH's attention because of trips caused by spurious signals on the power line.

F. Main Controller

Power subsystem main controller (PSSMC) has the task of coordinating the various operations in the power subsystem. An FPGA based controller is designed for VIKSAT1. The FPGA gets analog data via the analog to digital converters and digital status signals directly. It in turns controls the cell balancing circuits, charger on/off, load protection circuits and talks to C&DH.

GA based charging optimization of lithium-ion batteries in small satellites, (pdf 13 KB), Saurabh Jain

Small spacecraft that are powered by solar energy have limitations because of the size of their solar panels. The reasons for this are the size of the satellite itself, weight and cost issues. These issues are an inherent part of small satellite programs. With the limitations on the solar panel size, it is generally hard to comply with the demands from all the satellite subsystems, payloads and batteries at the same time. We confront similar issues with VIKSAT1, a small satellite prototype being developed at CSU. To overcome these problems we have developed and adopted a power management optimization scheme that runs in real time in the satellite. Though the scheme has been adapted to the specific needs of VIKSAT1 it is general enough to be suited to a wide variety of spacecraft, from small satellites to cruisers that travel distant places.

The proposed power management scheme primarily involves scheduling of loads (various subsystem operations, payload experimentation, battery charging, etc.) so that power utilization is at its optimum. At the same time it controls the amount and rate of charge that comes out of the batteries during eclipse and that goes into the batteries during the day time. This battery management becomes even more important with the use of Lithium-Ion batteries. These batteries have the advantage of high power density and light weight, but are very sensitive to over charging and over discharging. The power management program involves breaking everyday satellite activities into ‘tasks’ that may be periodically updated from the ground. A task is then fitted into a ‘template’ that identifies the ultimate ‘load’ for the power distribution system, which comprises the regulators to the subsystems. A ‘frame’ is generated for each load which then gets stamped with a priority index, start time, end time, modularity, interrupt ability, correlation identifiers, and other auxiliary information. The loads are then fed to a Genetic Algorithm (GA) based optimizer with the corresponding frame contents as the constraints to the GA. The output is a queue that has a schedule of operations in it. Since Genetic Algorithms are computationally intensive, we propose an FPGA based implementation of it. FPGA implementation allows a coarse parallelism and pipelining on the same platform. This speeds up the GA runs and makes them suitable for real time implementation.

On optimization of sensor selection for aircraft gas turbine engines, (pdf 957 KB), Ramgopal Mushini

Many science and management problems can be formulated as global optimization problems. Conventional optimization methods that make use of derivatives and gradients are not, in general, able to locate or identify the global optimum.

Sometimes these problems can be solved using exact methods like brute force. Unfortunately these methods become computationally intractable because of multidimensional search spaces. Hence the application of heuristics for a class of problems that incorporates knowledge about the problem helps solve optimization problems.

Sensor selection optimization can lead to significant improvements in thecontrollability and observability of a dynamic system. The aim of this research is to investigate optimal or alternate measurement sets for the problem of aircraft gas turbine engine health parameter estimation. The performance metric is defined as a function of the steady state error covariance and the cost of the selected sensors. A brute force search for the best sensor set is too computationally expensive. Therefore a probabilistic approach is used to perform a search for a near-optimal sensor set. In view of the need for global optimization of sensor selection for health parameter estimation, a genetic algorithm is also developed. A genetic approach will perform a search for a near-optimal sensor set. This will allow the health of the aircraft gas turbine engine to be estimated using fewer sensors while still obtaining an acceptable estimation error covariance, thereby minimizing the financial cost of the acceptable sensor set.

Particle swarm optimization applied to sensor selection for aircraft engines, (pdf 10KB), Chunming Yang

A New Particle Swarm Optimization (NPSO) method is proposed and comparisons are made to the traditional Particle Swarm Optimizer (PSO) invented by Kennedy and Eberhart in 1995. PSO is motivated by the social behavior of organisms, such as bird flocking and fish schooling. Each particle studies its own previous best solution to the optimization problem, and its group's previous best, and then adjusts its position (solution) accordingly.

The optimal value will be found by repeating this process. In the NPSO proposed here, each particle adjusts its position according to its own previous worst and its group's previous worst to find the optimal value. Simulation will show that NPSO converges faster than PSO in several benchmark functions.

Mathematical analysis will also show the advantages of NPSO over PSO. Aircraft engine sensor selection (for health parameter estimation) is an NP-complete problem. In this dissertation, PSO and/or NPSO will be applied to the problem, and advantages over other sensor selection approaches will be demonstrated.

Mailing Address
Cleveland State University
Embedded Control Systems Research Lab
2121 Euclid Avenue
SH-310
Cleveland, OH 44115-2214

Phone: 216.875.9670
Fax: 216.687.5405



Web Content Contact
Saurabh Jain

Web Technical Contact
Sunil Gavini