CONTROL VALVE SE LECTION: how big an issue?
By Pulp & Paper Canada
By Pulp & Paper Canada
Sub-standard control valve performance is probably the biggest contributor to process variabilityDoug Nelson, EnTech Control Inc., West Vancouver, BC; Bill Davey, Norpac Control Inc., Vancouver, BC. T…
Sub-standard control valve performance is probably the biggest contributor to process variability
Doug Nelson, EnTech Control Inc., West Vancouver, BC; Bill Davey, Norpac Control Inc., Vancouver, BC. This is a modified version of a paper given at the PacWest Conference, Whistler, BC, May 1999.
There is little doubt that as year 2000 approaches, the Canadian pulp and paper industry will “survive” in the 21st century. What is not known is what our products will look like, what our organizations will look like and what magnitude of change will be required to return the Canadian industry to a strong leadership position.
Global pulp and paper markets will remain extremely competitive with continual pressure on commodity pulp and newsprint prices. The industry has already responded with specialty value-added paper products and a constant search for “differentiation” in pulp beyond the traditional higher quality northern bleached softwood pulp.
Long-term prosperity will depend on our ability to continue to produce the highest quality product at the lowest possible cost. This is a huge challenge that we are just beginning to grasp. Re-engineering the organization, trades flexibility and contracting out of requirements for specialty expertise are already having an effect on profitability in many Canadian mills. This is the beginning of a cost cutting process that must culminate with minimum consumption of expensive chemicals, efficient production and optimized energy use and worldwide leadership in process efficiency, product quality and maintenance efficiency.
“Who cares about variability management?” Variability management programs offer major opportunities for improving business performance. Process variability directly affects the amount of chemical and energy consumed, operating efficiency and the quality of the product. Knowing the amount of variability in key processes and its effect on production and cost will mean that the most important process loops are optimized and the right equipment is fixed at the right time. This ultimately results in lower production costs and higher maintenance efficiency using the existing equipment. Who cares about variability management? We should all care.
Sub-standard control valve performance is the single biggest contributor to process variability. The reasons for this sub-standard performance include oversized valves, incorrect flow characteristics and valve tracking non-linearities.
In this paper, a case study (simulated typical stock refining system) approach has been used to demonstrate the effect of oversized control valves, incorrect valve characteristics and unacceptable valve dynamic performance on process control.
In the ideal case, control valve selection is based on accurate process data gathered over the entire operating range. The hydraulic characteristics of the system (a function of pump selection, pipeline design, fibre characteristics and others) are used to select the best flow characteristic to achieve relatively linear loop dynamics. Cavitation and noise issues are addressed. An actuation package is selected to ensure that the valve tracking non-linearities are minimized or at least appropriate for the application. Process simulation is used for valve selection on complex/interactive processes.
The reality in mills today is that more than 50% of all control valves are oversized and operate at less than 50% open at design conditions. This happens because incorrect pressures were used for selection, valves were selected to satisfy potential future requirements or process conditions have changed. Often, the appropriate valve characterization is not applied. Control valves often operate for long periods without maintenance and deteriorating dynamic performance results in increased process variability. Many mills do not recognize the effect of these deficiencies on process efficiency and operating cost.
Case study process
The process design (pump selection, pipeline design, control valve selection and other factors) contributes in a fundamental way to the hydraulic characteristics of the system and to the process dynamics of the control loops in the process system. The process design used in the simulated stock refining system, Fig. 1, consists of the following process and control equipment:
Goulds pump – 3175, 10x12x22, 17-in. impeller, 1180 rpm.
Consistency control at 4% using a 4-in. V-ball control valve.
Pressure control at 26 psig using a 6-in. V-ball control valve.
Flow control at 1500-3000 USgalpmin using an 8-in., 10-in. or 12-in. V-ball control valve.
14-in. pulp stock line with 10-in. recirculation header and 4-in. dilution line.
Supply tank level – 10 ft
Supply pump to discharge – 15 ft
A primary objective of this simulation is to profile the process dynamics of the control loops over the entire operating range and to define the effect of valve selection on process variability. The mechanical energy balance is solved to define available valve pressure drop at each flow condition. ISA standard equations are applied to calculate flow-rates. Valve dynamic blocks are used to determine the effect of valve tracking non-linearities on the process. The simulation development was facilitated with the EnTech SimTools software, which contains libraries of pump and valve performance curves, pipe friction loss estimates and valve dynamic blocks. Thus, the effect of control valve selection, pump selection and pipeline design are easily evaluated.
The effect of oversizing
The simulation was used to compare an 8-in., 10-in. and 12-in. V-Ball for the Flow control loop. Figure 2 plots the process gain (defined as ) over the entire flow range for each valve, and indicates the valve operating range. Selecting the 12-in. valve results in the highest process gain and the lowest valve operating range (32 to 48%). The 8-in. V-Ball results in the lowest process gain and the highest valve operating range (55 to 85%). It is interesting to note that the small operating range of the 12-in. V-ball results in a relatively constant Cv slope over the demand flow range. The process gain decreases at the higher flow rates, due to the lower available pressure drop. Conversely, the process gain of the 8-in. V-ball increases at the high flow rates due to the dramatic increase in the slope of the Cv curve at higher valve positions.
Table 1 summarizes typical variability results for the 8-in., 10-in. and 12-in. valves assuming a 1% control valve resolution. Selection of an 8-in. valve will result in substantially lower variability. This is because the amplitude of a flow limit cycle resulting from the 1% valve resolution is proportional to the effective process gain. In spite of this, an oversized 10-in. or 12-in. valve will often be found on this service with the significantly increased variability and potential for limit cycling.
The inherent flow characteristic of most valves used in the industry is equal percentage. This includes all V-ball and eccentric disc design valves. The equal percent characteristic is ideal for applications where increasing flow demand results in a decreasing valve pressure drop due to increased pipe friction and decreasing pump head. For systems where the valve inlet and outlet pressures are fairly constant, a linear characteristic should be considered to minimize the spread in the process gain over the operating range. Ideally, the process gain should not vary by more than a factor of 2 to enable a single set of controller tuning parameters to provide acceptable control performance over the demand range.
In the simulated refiner process system, the pressure valve inlet pressure is controlled at approximately 26 psig. The outlet pressure is relatively constant. Figure 3 shows the process gain and valve position over the demand flow range for a correctly sized and well-maintained 6-in. V-ball with inherent equal percent characteristics. Also shown is the process gain curve for the same valve with a linear cam in the positioner. Note that the range in
process gain for the standard equal percent characteristic unit is approximately 0.16 to 0.4 psi/ % output. Assuming that the pressure controller is tuned for average demand flow conditions, this loop will be very aggressive (and potentially cyclic) at high recirculation flows (low demand flows) and be very sluggish at low recirculation flows (high demand flows). Adding the linearizing cam reduces the range of process gain (0.23 to 0.39 psi/% out). This brings the gain within the Entech Control Valve Dynamic Specification recommendation and should enable tuning of the controller for reasonable control throughout the operating range.
Stable pressure control is important to both the consistency and flow control loops. Installing the correct cam or programming a custom characteristic in the positioner is a low cost solution to a potentially expensive problem.
Even new control valves can exhibit hysteresis, stiction and backlash totaling 5% and higher. This means it may take a controller output signal change of 5% or more to initiate a valve response to a demand change. There is no way to compensate for this valve performance deficiency in control strategy or tuning. To achieve an acceptable level of control, it is imperative that control valves be ordered and maintained to rigid performance specifications. More and more mills are now demanding compliance to EnTech’s Control Valve Dynamic Performance Specification. This specification requires that “the combined backlash and stiction is not to exceed 1.0% as installed during normal process operation.”
Figure 4 compares the pulp consistency variability exhibited with a 4% backlash and stiction valve vs. a replacement valve complying with the maximum 1% backlash and stiction per the EnTech specification. Note that replacing the under-performing valve with a well-maintained “spec” unit reduced the variability from an average 0.95% to 0.45%. The result is a more consistent fibre flow to the refiner and, ultimately, a more consistent product.
Survival in today’s competitive world markets depends on reducing cost and maximizing production. An active program to find and fix process control problems is a key ingredient.
This paper has shown that the ability of the process control system to achieve optimum control is dependent to a large degree on valve selection. Oversizing the control valve is a common problem, which results in increased process variability and potential for limit cycling. Fitting the control valve characteristic to the hydraulic characteristic of the process system will make the control loop dynamics more linear and will enable a single set of tuning parameters to provide optimum control over the range of operating conditions. This is especially important in process systems with interactive control loops where the success of the tuning strategy depends on uniform control dynamics. Finally, minimizing the valve tracking non-linearities will reduce vulnerability to limit cycling and compromised control performance. It is important to realize that all of the loops in the case study system interact, and that deficient performance of any one of these can affect the process variability of the other loops.
The first step in improving the performance of the process control system is to recognize the problems. There are some key indicators that signal something may be wrong with the control loop and costs are probably increasing. Is the loop in manual? Is the loop difficult to tune? Does the process variable cycle? Is the chemical consumption rate higher than expected? Is the process variable filtered to cover up a problem? A “Yes” answer to any of these questions signals a control problem. In many cases the control valve is the source of the problem, and the solution is simple and economical.
TABLE I. Effect of valve size on control resolution and variability.
|Valve size||Valve operating range||Resolution||Process gain||Process variability|
|(% of travel)||(1% steps)||(Kp)||(average)|