控制回路
一、控制回路基础 | Control Loop Fundamentals
二、PID控制 | PID Control
三、PID整定 | PID Tuning
四、高级控制策略 | Advanced Control Strategies
五、控制回路故障与诊断 | Loop Troubleshooting & Diagnostics
PID Tuning — Superheater Steam Temperature Control
Hans, we're having trouble with the superheater temperature loop — TIC-501. It's been oscillating with a period of about 6 minutes, amplitude ±15°C around the 540°C setpoint. The operators keep switching it to manual because the oscillation is causing the attemperator spray valve to hunt back and forth. Can you take a look?
Hans，过热器温度回路TIC-501有问题。一直在振荡、周期大约6分钟、 振幅在540°C设定值附近±15°C。操作工老把它切手动，因为振荡导致减温 器喷水阀来回摆动。你能看看吗？
±15°C on a superheater is too much — that's going to cause thermal fatigue in the header. Maria, can you give me the process characteristics on this loop?
过热器±15°C太过了——这会造成联箱热疲劳。Maria，能给我这个回路 的过程特性吗？
It's a slow loop. The dead time from spray valve to the temperature sensor is about 45 seconds — that's the transport delay through the attemperator and the pipe run. The time constant is around 90 seconds. So it's a dead-time-dominant process, which is always tricky for PID control.
这是个慢回路。喷水阀到温度传感器的死区时间大约45秒——这是通过 减温器和管段的传输延迟。时间常数约90秒。所以是死区主导过程，PID 控制一向棘手。
Dead-time-dominant — that explains the oscillation. Tom, pull up the current PID parameters for TIC-501.
死区主导——这就解释振荡了。Tom，调出TIC-501当前PID参数。
Kp = 2.5, Ti = 120 seconds, Td = 0 (derivative is turned off). The operator told me the previous engineer set these based on "experience" — no formal tuning was done.
Kp=2.5、Ti=120秒、Td=0（微分关了）。操作员告诉我前任工程师 是凭"经验"设的——没做过正式整定。
That Kp is way too aggressive for a loop with 45 seconds of dead time. And with Ti at 120 seconds, the integral action is fighting the process dynamics. Let me run a step test — Tom, put the loop in manual and hold the output at 50%. Wait for steady state... OK. Now step the output to 55% — a 5% bump.
那个Kp对有45秒死区时间的回路来说太激进了。Ti 120秒、积分作用 在和过程动态打架。我来做阶跃测试——Tom，回路切手动、输出保持50%。 等稳态……好。现在输出阶跃到55%——5%阶跃。
(watching the trend) Temperature starting to rise... slowly. After 45 seconds — there's the first response. It's still rising... leveling off at a 12°C increase. Process gain is about 2.4°C per percent output.
（看趋势）温度开始上升……缓慢。45秒后——终于有响应了。还在 升……稳定在升温12°C。过程增益大约每1%输出2.4°C。
Process gain: 2.4 °C/%. Dead time: 45 sec. Time constant: about 90 sec. Now I can calculate the tuning parameters. For a dead-time-dominant process, I'll use Lambda tuning with Lambda = dead time (conservative approach since this is a critical loop).
过程增益：2.4°C/%。死区时间：45秒。时间常数：约90秒。现在我能 算整定参数了。对死区主导过程，用Lambda整定、Lambda取死区时间 （保守策略，因为这是关键回路）。
Kp = Time constant / (Process gain × (Dead time + Lambda)) = 90 / (2.4 × (45 + 45)) = 90 / 216 = 0.42 Ti = Time constant = 90 seconds. Td = 0 (don't use derivative on noisy temperature signals). Tom, update TIC-501 with Kp = 0.42, Ti = 90 sec, Td = 0.
Kp = 时间常数 / (过程增益 × (死区+Lambda)) = 90 / (2.4 × (45 + 45)) = 90 / 216 = 0.42 Ti = 时间常数 = 90秒。Td = 0（噪声大的温度信号不用微分）。 Tom，更新TIC-501参数：Kp=0.42、Ti=90秒、Td=0。
Updated and switched to AUTO. Paul, watch the trend for the next 15 minutes.
已更新并切到AUTO。Paul，观察趋势15分钟。
(8 minutes later) It settled at setpoint within 90 seconds with one small overshoot of about 4°C. Now it's holding steady at 540 ± 2°C. That's a beautiful response. The operators are going to love you, Hans!
（8分钟后）90秒内稳定在设定值、一次小超调约4°C。现在稳定在 540±2°C。漂亮的响应。操作工们得爱死你，Hans！
Control Valve Stiction Diagnosis — Cooling Water Loop
Anna, the cooling water temperature loop — TIC-702 — is behaving strangely. The controller output is cycling between 32% and 38% every 3 minutes, but the valve position feedback shows it's barely moving — it sticks at 34% for a while, then suddenly jumps to 37%, then back. The temperature is hunting ±3°C as a result. I think the valve has stiction.
Anna，冷却水温度回路TIC-702表现怪异。控制器输出每3分钟在32% 到38%之间来回，但阀门位置反馈显示它基本不动——卡在34%好一会儿、 突然跳到37%、又跳回来。结果温度在±3°C波动。我觉得阀门有粘滞。
Classic stiction pattern. When the controller output changes slowly, the valve doesn't respond until the pneumatic force overcomes the static friction. Then it jumps — "slip-stick" behavior. Lee, let's go to the valve and run some diagnostics. David, can you put TIC-702 in manual?
典型的粘滞特征。控制器输出缓慢变化时，阀门不响应直到气动力 克服静摩擦。然后突然跳——"滑-粘"行为。Lee，我们去阀门那做诊断。 David，能把TIC-702切手动吗？
(at the valve, opening the positioner cover) This is a Fisher DVC6200 digital positioner on a globe valve, 6-inch, Class 300. Let me check the diagnostics in the positioner.
（在阀门处，打开定位器罩子）这是Fisher DVC6200数字定位器配 截止阀、6英寸、300磅级。我查定位器内部诊断。
The positioner is logging "stick-slip detected" events — confirmed. The I/P converter output is oscillating because it's trying to overcome the friction. Supply air pressure is 4.2 bar — normal. No air leaks. Let me check the packing friction.
定位器记录有"检测到粘滑"事件——确认了。I/P转换器输出在振荡因为 它在努力克服摩擦。气源压力4.2 bar——正常。无漏气。我检查填料摩擦。
Lee, try a partial stroke test — cycle the valve 10% around the 35% position. Watch the position feedback versus the positioner demand signal.
Lee，做个部分行程测试——在35%位置附近让阀门摆动10%。看位置 反馈和定位器指令信号的对比。
Running... the demand signal moves smoothly, but the actual position lags, then catches up in a jump. The positioner is increasing the output pressure to the actuator to overcome the packing friction, but the friction is too high. The packing is the problem — it's over-tightened.
运行中……指令信号平稳移动，但实际位置滞后、然后突然跳到位。 定位器在增加去执行机构的输出压力来克服填料摩擦，但摩擦太大。 填料是问题——拧太紧了。
Let's loosen the packing gland bolts just enough to reduce the friction while maintaining a seal. Use the torque specification from the maintenance manual — 35 N·m on the gland follower nuts. Not more, not less.
我们把填料压盖螺栓松一点，刚好减少摩擦又能保持密封。用维护 手册的扭矩规定——压盖从动螺母35 N·m。不多不少。
(adjusting the gland bolts with a torque wrench) Done. Let me re-run the partial stroke test... much smoother now. The position feedback tracks the demand signal within 0.3% — no more stick-slip. The packing is still sealed — no leakage.
（用扭矩扳手调整压盖螺栓）好了。我再跑一次部分行程测试…… 现在顺滑多了。位置反馈跟踪指令信号误差0.3%以内——不再粘滑了。 填料仍密封——无泄漏。
David, we've fixed the valve stiction — it was over- tightened packing. Lee re-torqued to spec. You can put TIC-702 back in AUTO. You should see much better control now.
David，阀门粘滞已修好——是填料过紧。Lee按规范重新拧紧了。 TIC-702可以切回AUTO了。现在控制效果应该好很多。
(back in the control room) TIC-702 in AUTO — temperature is rock solid at setpoint, the valve output is steady, no hunting. Another one for the "valve stiction was the problem" file. Anna, I think 80% of our control loop issues are valve-related.
（回到控制室）TIC-702 AUTO模式——温度死死贴在设定值上，阀门 输出稳定、不再摆动。又一条"阀门粘滞是元凶"的案例。Anna，我觉得 80%的控制回路问题都是阀门相关的。