Comment

1 June 2005

Managing industrial power

How can a co-generation plant produce power cheaper than a large modern utility?

By Jim Shriver

Process industries are hooked on electrical energy.

Typically, in mills where large amounts of steam are required for the process, steam turbines are for co-generation of electricity. However, since the amount of co-generated power in most mills is insufficient to satisfy the mills total electric load, the mill purchases the balance of its requirements from the local electric utility as well as relying on them for backup service in case of generating equipment outages.

Electricity rates charged to industrial and commercial customers vary widely across the country. A recent survey comparing the monthly cost of 10,000 kWh (a small commercial user's consumption) indicated a wide variation between the highest ($1,270) and lowest ($186) cost.

This large variation is from the cost of the fuel used to generate the power by the providing utility. Access to extensive hydroelectric generation facilities tend to stabilize and reduce that area's electric costs, while urban area's rates have risen rapidly with the cost of fuel oil.

Thus, a customer's incentive for requiring a power management solution depends heavily on his specific location and specific electric power economics. Real time pricing contracts swing drastically based on supply and demand.

What makes sense on the night shift may not apply to the day shift for example. Weekend and summer rates may require a different operating procedure.

Regardless of the specific charges assessed by the utility, customer service charges usually have two components:

1) An energy charge that costs so much per kilowatt-hour used during the billing period. This is usually subject to a sliding scale with the rate per kWh decreasing with increasing usage. The energy charge may also be broken down further with a fuel adjustment charge. The fuel adjustment charge is a convenient way for the utility to modify its rates with little or no regulatory delay in times of rapidly escalating fuel costs. The energy charge may also have other components, but the net result is an overall rate, which applies to each kWh used.

2) A demand charge at so much per kilowatt of demand, where demand is the highest kW load imposed on the tie line usually averaged over a 15-, 30-, or 60-minute interval during billing period (or the highest demand period established during the month). The theoretical basis of the demand charge is it pays the utility for the capital cost of generating equipment capability. Some service contracts contain a "ratchet" clause, which specifies that once a customer establishes a monthly demand higher than the previous demand, the new demand charge is assessed for a year or more even if ensuing monthly demands are reduced. This ratchet provision obviously gives the customer a strong incentive to closely control his tie line demand. The demand charges may also be on a sliding scale, decreasing with increasing demand.

Demand charges vary greatly typically ranging from $6.00 to $30.00 for each kilowatt of demand. Even though the demand charge component of a bill may be a large amount, its proportion of the total bill may be small. For example, consider a mill whose highest demand is 20 megawatts and an average load of 16 megawatts. Rates are assumed $.05/kWh energy, $10/kW demand.

kWh usage = 16,000 kW * 24h/d * 30 d/m = 11,520,000 kWh
So, the energy charge is 11,520,000 * $.05 = $576,000
And the demand charge is 20,000 * 10 (80% LF) = $200,000
For a total of $776,000.

In this example, a 3 MW (15%) reduction in demand would yield a $360,000/year savings. Lowering the demand, however, means increasing the Load Factor (LF)—the ratio of the average kW load to the highest load. The initial LF in the example above was 80% (16/20). Reducing the demand from 20 to 17 MW increases the LF to 94% (16/17). The minimum possible monthly demand charge in the example above would be the situation where the customer's tie line load was exactly constant throughout the billing period, i.e., 100% load factor (16MW average/16MW demand).

A customer can control his load factor by:

1) Scheduling and/or sequencing of process loads to shed and restore the loads at appropriate times.

2) Controlling the amount of power co-generated.

The first alternative is usually very difficult to accomplish in a mill with continuous processes. The exception is the situation where one can establish inventories at points within the process. For example, a large electric motor driven wood chipper in a pulp mill, which has excess capacity, can shut down for periods without affecting the rest of the mill. Load shedding is popular in commercial installations where air-conditioning loads and the like may selectively shut off on a priority basis by simple and inexpensive control devices.

The second alternative, controlling the amount of co-generated power, is the area of most interest to co-generating mills, not only because it usually has little or no effect on the process side of the mill, but also because internal political problems do not arise. Process operators usually have little incentive or interest (and are sometimes hostile) towards the powerhouse operator controlling their equipment.

Co-generation economics

There are essentially three types of steam turbine generators at work in power generation: backpressure turbines, condensing turbines, and extraction turbines.

There are a multitude of combinations of these turbine types including backpressure units with extractions and units with multiple extractions with or without condensing stages.

How can a co-generation plant produce power cheaper than a large modern central station utility? Comparing heat utilization in a utility plant versus a co-generation plant can see the answer.

The thermal efficiency of a utility plant can be about 39% or a heat rate of @ 8750 Btu/kWh. Adding transmission and distribution losses of 5%, the net output to customers' costs is about 9200 Btu/kWh.

However, in an industrial plant where power generates from a non-condensing turbine, the heat in the low-pressure steam can also go for process use. In fact, heat utilization in an industrial plant can be, up to 85%. Thus, the ratio of heat utilization in an industrial plant to that of the utility is at least 2 to 1.

However, in general, the cost of power produced by condensing turbines in an industrial plant is higher than the incremental cost of power from the utility. Only in cases where the utility rates are high and the industrial plant's incremental fuel cost is extremely low does power produced via condensing turbines cost less than utility generated power.

Condensing power is expensive because of the large proportion of the heat in the throttle steam is lost in the condenser. Ideally, condensing turbines should not run at all, or if they must be for some reason, their loading should be at a minimum.

If the unit running in an industrial situation is a backpressure turbine exhausting into a low-pressure header, the process can use all the heat of the low-pressure steam.

The heat chargeable to the production of electricity is strictly the difference between turbine inlet and exhaust enthalpy.

In the condensing turbine example, not only was the enthalpy difference counting as cost in the production of electricity, but also the large amount lost to the environment in the condenser.

Turbine dispatch strategies

There are three control functions involved in turbine dispatch applications.

Tie line demand control is to control the load on the tie line below the established demand limit.

Header pressure control maintains the steam balance in the low-pressure headers and minimizes the use of pressure relief valve (PRV) stations.

Load allocation is to maximize the co-generation of electricity by allocating load and flow to the most efficient turbine sections.

These functions are in order of decreasing priority. Most plants can sustain excursions in header pressures should tie line demand control require it for short periods. For example, if tie line demand is too high and the condensing turbine is fully loaded, venting steam from the low-pressure header can divert to generate additional power from an extraction turbine. Similarly, shifts in process steam loads will cause changes in extraction flows and may drive load allocation away from optimum. It is important to establish priorities in the face of somewhat conflicting interacting objectives. In all cases, the driving force is the specific economics involved at the plant site.

Tie line demand control

There are three fundamental functions to a tie line control system: input measurements, calculations/logic, and output signals.

Tie line demand signal inputs usually are pulse type with each pulse representing so many kilowatt-hours used. Sometimes analog signals come as well, and they represent tie line load in megawatts. The start (or end) of each tie line demand interval can either be defined by a contact input from the metering equipment or by the control system via an internal clock. Ideally, the control system should use the same metering and timing signals used by the power company for billing purposes.

Other measurements required include measurements of kilowatt loading and steam flows on each turbine to be controlled.

The calculations and logic function analyzes the input data and determines what control action should be taken. The system integrates the kilowatt-hour input signals from the start of each demand interval. At any point in the interval, the allowable tie line load (in kilowatts) is the slope of a line drawn from the current accumulation point to the target usage at the end of the period.

Note the system controls demand in kilowatt-hours used over a period of time or interval. Instantaneous tie line load in kilowatts at times may exceed the kilowatt demand limit with no penalty incurred, but the integrated period usage in kilowatt-hours will not be allowed to exceed the allowable interval usage limit.

The system then compares the present instantaneous (kW) load on the tie line with the computed allowable. The system will output appropriate signals (usually increase/decrease pulses) to the turbine controls to drive the turbine loading in the appropriate direction.

Logic limits turbine loading to operator entered minimum/maximum limits. If multiple turbines are to be driven, the system picks up load on the most efficient turbine and conversely reduce loading on the least efficient.

Logic is also required to recognize the situation where tie line demand is low and the turbines are at low limits. There is no cost penalty for under usage of the tie line as long as condensing flows are at their minimum levels.

The system may also be designed to open a vent valve on the low-pressure header should tie line demand be high and the condensing turbines fully loaded. This action increases the power generated by the extraction turbines without upsetting severely the header balances.

The on board logic ensures sufficient boiler and extraction turbine capacity is available before taking this action.

Interface with existing turbine controls usually is done by providing increase/decrease pulse outputs (thru either timed contact outputs or pulse trains) to the speed/load motor-driven potentiometer in the turbine electro hydraulic control (EHC) system. This device provides a means for varying turbine speed/load demand and/or extraction pressure set points within the EHC system from a remote device or location. The operator can manually operate this device as well via increase/decrease pushbutton on the EHC panel. This EHC device determines the characteristics of the output signal. Modern Turbine Controls will interface seamlessly into the plant wide control system in a number of ways that are usually "transparent" to the customer. For example, General Electric is a major supplier of gas and steam turbines.

The Mark Five (MK V) TMR Machine Controller is a self contained, highly secure control system. MK V provides bi-directional communication with external systems (like the Foxboro I/A Series). GE can provide either "I" processors that communicate via Modbus or "G" processors that communicate via Ethernet. Single or redundant communication processors are readily available. The processors for multiple turbines can interconnect with an ARC Net network on the GE side. Most other suppliers of turbines have standardized on Modbus as the standard connection protocol to DCS systems.

Header pressure control

Steam headers are actually low capacity reservoirs of energy distributed through the plant. Flows are continually entering and leaving through various paths and at changing rates. A balance of inlet and outlet flows is only possible by the action of a pressure controller.

When demand exceeds supply, the controller must deliver more steam from a higher-pressure source, usually the next higher pressure header. When supply exceeds demand, pressure rises, and steam may spill over into the atmosphere or into a lower pressure header.

Header pressures are usually under the control of extraction or backpressure type turbines, with the actual control done by the EHC, which works reasonably well. However, in plants where multiple turbines extract or exhaust to the same header, usually only one turbine controls header pressure while the others are base loaded manually by the operator.

This is fine during steady-state conditions, but during severe steam load upsets, the controlling turbine may not have the swing capacity to restore the steam balance quickly. Vent valves and PRV stations then operate sometimes for long periods before conditions stabilize.

By operating the base loaded turbines, the turbine dispatch system can restore the steam balance faster and minimize or eliminate the usage of the vent valves and PRV stations. In one paper mill application, total daily flows thru these valves went down by a factor of three.

In addition, as process steam loads change, the turbine dispatch system can ensure that the turbines controlling header pressure stay within their normal control range.

Floating header pressure control can be a part of the turbine-dispatch control system too. This function maximizes co-generation by operating the low-pressure header at the minimum level, which satisfies all users.

For instance, in a system that contains eight control valves, all the positions of the eight control valves of low-pressure steam users match up against the highest. A valve position controller then attempts to maintain the most open valve at 90-95% open, by driving the set point of the extraction pressure controller in the required direction.

The valve position controller acts very slowly to minimize upsets to users. The set point of the PRV between the high and low-pressure headers works at the same time to prevent it from overriding the extraction pressure controller.

Turbine load allocation

The purpose of this function is to maximize co-generated power by optimizing the load allocation of process steam among all turbine sections. This function can only take place if there are multiple turbines extracting or exhausting into the same header and the capacity of the turbines normally exceeds the process demand. Optimizing the load allocation of process steam among the turbine sections will result in more power co-generated for a given process steam demand.

For example, consider a simple two-turbine system both exhausting into the same header. The relationship between exhaust enthalpy and flow is such that as flow increases exhaust enthalpy decreases. Here are all of the possible loading allocations for this two-turbine system with contours of constant electrical production plotted for various loading arrangements.

The process steam demand is the dotted, diagonal line. The possible loading allocations must also fall on the diagonal line. Note that at either extreme of the diagonal line is where electrical power is the most.

Simple, equal loading of the turbines provides for minimal optimization. Should there be efficiency differences between the turbines in the above example, the turbine with the highest efficiency should be loaded to the maximum extent.

For a simple turbine system, a turbine-load allocation strategy could happen with manual control. However, as the number of turbines and extraction points increase, the problem quickly becomes too complex for manual optimization.

The advances in computer memory and storage capacities are what enable advanced strategies to work in a cost effective manner that previously would have been too expensive.

Behind the byline

Jim Shriver is an industry consultant in the power division of Invensys Process Systems. He spent 25 years at Foxboro developing their power tune-up concept.