坎貝爾圖
一種頻率對轉速的圖表,揭示旋轉機械中的臨界轉速、陀螺效應分裂與共振危險區——從微型渦輪機到多兆瓦壓縮機組。
定義
A 坎貝爾圖 (亦稱為 進動轉速圖 或 干擾圖)是一種繪製 自然頻率 of a rotor-bearing system on the vertical axis against rotational speed on the horizontal axis. Diagonal excitation-order lines (1×, 2×, 3×…) are superimposed; wherever an excitation line crosses a natural-frequency curve, a 臨界轉速 exists. The diagram is the primary tool for determining whether a machine's operating range is safely separated from 共振 條件的圖表。
簡言之:坎貝爾圖回答了一個問題—— "At which speeds will this rotor resonate, and how close are those speeds to where I plan to operate?"
歷史背景
Wilfred Campbell 於 1924 年在通用電氣研究汽輪機盤的周向波時發表了此概念。他最初的圖表繪製了盤的振動模態對轉速的關係,以預測運轉期間破壞性共振出現的位置。
The approach filled a gap that had troubled engineers since the 1890s. W. J. M. Rankine's 1869 shaft-whirling analysis had incorrectly predicted that supercritical operation was impossible. Gustaf de Laval proved otherwise by running a steam turbine above its first critical speed in 1889. Henry Jeffcott's landmark 1919 paper finally explained 為何 supercritical operation is stable, but Campbell's diagram gave engineers the 視覺化工具 以精確預測這些危險轉速的位置——以及如何設計避開它們。
在隨後的幾十年中,此概念從盤振動擴展至完整的橫向轉子分析、扭振分析,甚至聲學。如今,所有主要的 API、ISO 與 IEC 旋轉機械標準皆要求或建議進行坎貝爾圖分析。
圖表結構解析
坎貝爾圖在單一圖表中包含四類資訊。正確判讀交點前,必須理解每一層資訊。
座標軸
橫軸為轉速,通常以 RPM 或 Hz 表示。縱軸為頻率,以 Hz 或 CPM 表示。當兩軸使用相同單位時,1× 激振線恰好呈 45°——這是驗證比例尺是否正確的有效視覺檢查方法。
自然頻率曲線
Each curve represents one vibration mode of the rotor-bearing-support system. In the simplest case (rigid bearings, no gyroscopic effects), these curves are horizontal lines because the natural frequencies do not change with speed. In reality, gyroscopic moments and speed-dependent bearing stiffness cause the curves to slope, split, or both.
Modes are labeled by deflection shape: first bending (one antinode), second bending (two antinodes with one node), third bending, and so on. Torsional and axial modes may also be plotted if relevant.
正向與反向進動
當陀螺效應顯著時,每個非旋轉狀態下的自然頻率會隨著轉速增加分裂為兩條曲線:
- Forward whirl (FW): 模態進動方向與軸旋轉方向相同。陀螺剛化效應使其頻率 返回頂部.
- 反向進動(BW): 模態進動方向與旋轉方向相反。陀螺軟化效應使其頻率 降低.
正向進動模態是 不平衡驅動共振的主要關注點,因為不平衡會激勵同步正向進動。
激振階次線
這些是從原點發射的直線對角線。每條線代表一個激振,其頻率為轉速的固定倍數:
| 線 | 關係 | 典型來源 |
|---|---|---|
| 1× | f = 1 × RPM/60 | 質量不平衡,轉子彎曲 |
| 2× | f = 2 × RPM/60 | 不對心,軸裂紋、橢圓度 |
| 3×、4×… | f = n × RPM/60 | Gear mesh, vane/blade pass, coupling defects |
| 0.43–0.48× | f ≈ 0.45 × RPM/60 | Oil whirl in fluid-film bearings |
| 葉片通過頻率 | f = Z × RPM/60 | 葉片數 Z × 運轉速度 |
交點 = 臨界轉速
激振線與自然頻率曲線之間的每個交點都標示了一個潛在的共振點。該交點處的 RPM 值即為該特定模態-激振組合的臨界轉速。如果運轉範圍包含或接近該 RPM,機器就有產生高振動幅值的風險。
互動式坎貝爾圖
下方的 SVG 顯示了一個雙軸承、柔性軸轉子的典型坎貝爾圖。將游標懸停在元素上以識別模態、激振線和臨界轉速交點。
圖 1 — 柔性雙軸承轉子的坎貝爾圖。金色圓圈標示臨界轉速 (CS₁, CS₂)。琥珀色帶顯示運轉速度範圍 9,000–12,000 RPM。
如何閱讀與解讀坎貝爾圖
逐步解讀程序
識別運轉速度範圍
找到標示最小和最大連續運轉速度的垂直帶或刻度線。在圖 1 中,此範圍為 9,000–12,000 RPM。
首先追蹤 1× 線
1× 同步線最為關鍵,因為不平衡(存在於每個轉子中)會在 1× 運轉速度下產生激振。找出它與順向進動曲線交叉的每個點。
讀取交點的水平座標
Each intersection's x-coordinate is a critical speed. Record each one along with the mode number it involves.
檢查 2× 及更高階的交點
對 2×、3×、葉片通過頻率及次同步線重複此步驟。這些交點是次要臨界轉速——能量低於 1×,但仍可能引起振動問題,特別是當激振源強烈時。
計算分離裕度
對於每個臨界轉速,計算其到運轉範圍最近邊緣的百分比距離。並與適用的標準 (API 617, API 612, ISO, OEM 規格) 進行比較。
評估曲線斜率
陡峭向上傾斜的順向進動 (FW) 曲線表示強烈的陀螺效應——常見於懸臂轉子。近乎平坦的曲線則表示系統由軸承剛性主導。
識別危險區域
如果兩個臨界轉速夾住運轉範圍且裕度不足,則必須修改設計:軸承剛性、軸直徑、支撐剛性或運轉速度必須改變。
⚠️ 常見的誤解: 逆向進動模態很少對不平衡激振產生響應,因為不平衡僅產生順向進動。與逆向進動 (BW) 曲線的交點通常不是真正的運轉臨界轉速——它們被包含在圖中是為了完整性,以及存在其他激振源的情況(例如密封中的反向流動)。
理解分離裕度
安全運轉要求運轉速度範圍必須遠離每個臨界轉速,以便共振放大在可容忍範圍內。所需的裕度取決於共振峰的尖銳程度,由 放大係數 (AF).
- 低 AF (< 2.5) means heavy damping — the rotor can operate close to or even at the critical speed without excessive vibration.
- 高 AF (> 8) 意味著尖銳的峰值——即使與臨界轉速偏離幾個百分點,也會導致危險的幅值增長。
典型的工業實踐要求 15–30% 的分離裕度,但具體要求取決於 governing 標準和 AF 值。
陀螺效應與頻率分裂
當旋轉盤進動(晃動)時,會產生耦合兩個垂直平面運動的陀螺力矩。這種耦合將零轉速時的單一自然頻率分裂為任何非零轉速下的兩個不同頻率。
物理原理
具有陀螺效應的轉子運動方程形式如下:
其中 M 是質量矩陣, C 阻尼矩陣, G 斜對稱陀螺矩陣(與轉速 Ω 成正比),以及 K 剛性矩陣。由於 G 取決於轉速,特徵值——因此自然頻率——會隨 Ω 變化。
何者決定分裂幅度?
極慣性矩 (Ip) 與徑向慣性矩 (Id) 的比值控制著陀螺效應的強度。盤狀元件 (Ip/Id > 1) 產生強烈的分裂。細長軸段 (Ip/Id ≈ 0) 產生的分裂可忽略不計。
Overhung rotors (single-stage pump impellers, turbocharger wheels, cantilevered grinding wheels) exhibit the most pronounced gyroscopic splitting. In these designs, the forward-whirl first critical speed can be 20–40% higher than the zero-speed natural frequency, meaning the Campbell diagram differs dramatically from a simple "flat-line" model. Running a flat-line analysis for an overhung rotor will underpredict the first FW critical and overpredict the first BW critical, potentially leading to incorrect operating-speed decisions.
軸承類型如何影響坎貝爾圖形狀
軸承將轉子連接到定子,並定義決定自然頻率的邊界條件。不同的軸承技術會產生根本不同的圖形形狀。
| 軸承類型 | 剛性行為 | 對坎貝爾曲線的影響 | 其他注意事項 |
|---|---|---|---|
| 滾動軸承 (球軸承、滾子軸承) | 隨轉速幾乎保持恆定 | 自然頻率曲線大致平坦(水平),除非陀螺效應佔主導地位 | 缺陷頻率 (BPFO、BPFI、BSF) 會在非整數階次增加激振線 |
| Fluid-Film (Journal) | 剛性與阻尼隨轉速增加(索默費爾德數變化) | 曲線向上傾斜的陡峭程度超過僅由陀螺效應產生的程度 | 交叉耦合剛性可能導致不穩定(油膜渦動/油膜振盪);增加 0.43–0.48× 次同步線 |
| 傾斜瓦滑動軸承 | 剛性隨轉速增加;交叉耦合極小 | 與普通滑動軸承斜率相似,但穩定性更佳 | 依 API 617 標準,為高速壓縮機首選 |
| 主動磁浮軸承 | 透過控制演算法可程式化;可為恆定、增加或適應性 | 曲線可刻意塑形,將臨界速度移離運轉範圍 | 控制迴路頻寬限制高頻率下可達到的最大剛性 |
| 氣體軸承(箔片/氣靜壓) | 剛性隨轉速急劇增加;阻尼極低 | 曲線急劇上升;高 Q 值共振 | 低阻尼使分離裕度更為關鍵 |
各向異性支撐
當軸承支撐座或基礎在水平與垂直方向具有不同剛性時,每個模態會進一步分裂為水平與垂直變體。坎貝爾圖則會顯示更多曲線——每個模態包含水平前向進動、垂直前向進動、水平後向進動與垂直後向進動。這在具有柔性基礎的水平機器中很常見。
API 617 與分離裕度要求
對於石油、化學與氣體服務中的離心與軸流壓縮機,API 標準 617(第 8 版,2014 年;第 9 版,2022 年)強制要求進行嚴格的坎貝爾圖分析,作為橫向轉子動力學研究的一部分。
API 617 分離裕度公式
其中 SM 為所需分離裕度 (%),而 AF 為該臨界速度下不平衡響應(波特)圖的放大係數。
| AF 值 | 依公式之 SM | 解讀 |
|---|---|---|
| < 2.5 | 無需 SM | 臨界阻尼;可在臨界速度下運轉 |
| 3.5 | 8.5% | 中等阻尼;小裕度即可 |
| 5.0 | 12.1% | 典型用於傾斜瓦軸承 |
| 8.0 | 14.4% | 尖銳峰值;需要較大裕度 |
| 12.0 | 15.4% | 非常尖銳;接近 16% 上限 |
| > ~11 | ≤ 16%(上限) | API 將低於最低轉速的臨界轉速 SM 上限設為 16% |
將此應用於坎貝爾圖
在設計審查期間,工程師從坎貝爾圖讀取每個臨界轉速,然後檢查波德圖中對應的 AF。如果 SMactual ≥ SM所需,則設計通過。否則,工程師必須修改軸承、軸幾何形狀或運轉範圍,直到滿足所有裕度要求。
其他具有類似要求的標準: API 612(汽輪機)、API 613(齒輪箱)、API 672(成套空氣壓縮機)、ISO 10814(臨界轉速接近度公差)、ISO 22266(往復式機械的機械振動)。每個標準使用略有不同的公式或固定百分比閾值,但都依賴坎貝爾圖作為原始資料。
建立坎貝爾圖:分析與實驗
分析(有限元分析 / 轉移矩陣)方法
建立轉子模型
將軸、盤、葉輪、聯軸器和襯套離散化為樑單元(Timoshenko 或 Euler-Bernoulli)或 3D 實體/殼單元。包含質量、剛性和陀螺項。
定義軸承特性
輸入與轉速相關的剛性和阻尼係數(每個液膜軸承 8 個係數:Kxx,Kxy,Kyx,Kyy,Cxx,Cxy,Cyx,Cyy)。對於滾動軸承,使用恆定剛性值。
設定轉速範圍和增量
定義從 0 到至少最大連續轉速 115% 的轉速掃描(根據 API 617 跳脫轉速要求),並使用足夠細的 RPM 增量(通常為 100–500 RPM 步進)以準確捕捉曲線形狀。
求解複數特徵值問題
在每個轉速步進,求解 det(K + iΩG − ω²M) = 0 以找到自然頻率 ωn (虛部)和阻尼(實部)。虛部成為坎貝爾圖上的 y 座標。
繪製並疊加激振線
繪製所有模態與轉速的關係,添加 1×、2× 和其他相關激振線,並標記交點。
實驗方法(來自現場資料)
當機器已存在時,可以從升速或惰轉降速過程中的振動測量中提取坎貝爾圖:
- Mount accelerometers or proximity probes at bearing locations.
- 在緩慢啟動(或跳脫後惰轉降速)期間連續記錄振動。
- 生成 瀑布(級聯)圖:在連續 RPM 值下拍攝的 FFT 頻譜堆疊。
- 識別每個 RPM 切片中的頻率峰值——這些是由主導階次激發的自然頻率。
- 繪製峰值頻率與 RPM 的關係以產生實驗坎貝爾圖。
惰轉降速測試通常比啟動產生更清晰的資料,因為機器平順減速,沒有馬達啟動時的扭矩波動。從跳脫轉速到靜止進行惰轉降速,並進行連續高解析度資料採集(≥ 4,096 條線,0.5 秒平均)。如果機器使用變頻器,請編程 50–100 RPM/秒的線性斜坡以獲得最佳頻率解析度。
依機器類型分類之應用
| 機器 | 典型轉速範圍 | 坎貝爾圖關鍵關注點 | governing 標準 |
|---|---|---|---|
| 離心壓縮機 | 3,000–60,000 RPM | 多個臨界轉速;液膜軸承不穩定;密封交叉耦合;通常跳脫轉速以下有 2–4 個模態 | API 617 |
| 汽輪機 | 3,000–15,000 RPM | 葉片通過激振;暖機期間熱彎曲改變模態;高階次的盤模態 | API 612 |
| 燃氣渦輪機 | 3,600–30,000 RPM | Dual-spool designs require separate Campbell diagrams for each spool; squeeze-film damper effects | API 616 / OEM |
| 電動機 / 發電機 | 750–36,000 RPM | 2× 電源頻率的電磁激振;VFD 驅動馬達需掃過共振區 | API 541 / IEC 60034 |
| 泵 | 1,000–12,000 RPM | 懸臂葉輪具有強烈的陀螺效應;葉片通過激振;耐磨環剛性隨時間變化 | API 610 |
| 機床主軸 | 5,000–60,000+ RPM | 預壓角接觸軸承;高轉速下預壓損失導致頻率軟化 | ISO 15641 / OEM |
| 渦輪增壓器 | 30,000–300,000 RPM | 浮動環軸承具有複雜的內/外油膜動力學;次同步渦動常見 | OEM / SAE |
| 風力渦輪機變速箱 | 10–20 RPM(轉子);高達 1,800 RPM(高速軸) | 用於齒輪嚙合共振的扭轉坎貝爾圖;多種速比 | IEC 61400 / AGMA |
設計階段用途
在設計階段,坎貝爾圖指導關於軸徑、軸承位置、軸承類型以及葉輪/盤幾何形狀的決策。將臨界轉速移動僅 10% 可能需要將軸承跨距改變 50 mm 或軸徑改變 5 mm —— 該圖表向工程師精確顯示所需的偏移量。
故障排除用途
If a machine develops high 1× vibration at a specific speed, the Campbell diagram quickly shows whether that speed coincides with a predicted critical. If it does, the solution is either to change the operating speed, add damping (e.g., squeeze-film damper), or improve balancing quality. If it does not, the high vibration likely has a different root cause such as mechanical looseness or bearing defect.
運轉指導
坎貝爾圖定義了 禁止轉速範圍 —— 由於臨界轉速落在該帶域內而不允許持續運轉的 RPM 帶域。變速機(VFD 驅動壓縮機、負載跟隨的渦輪發電機組)必須審查其坎貝爾圖,以確保沒有連續運轉工況點位於禁止帶域內。在啟動或停機期間瞬態通過臨界轉速是可以接受的,前提是加速度足夠高以防止振幅累積。
測量圖表預測的內容
Balanset-1A 便攜式分析儀記錄您進行實驗坎貝爾圖所需的振動資料 —— 升速和惰轉降速期間的頻譜與 RPM 關係。現場雙平面動平衡校正。售價 €1,975 起。
相關圖表與曲線圖
坎貝爾圖是轉子動力學分析中幾個相互關聯的視覺化工具之一。每個工具都有獨特的用途。
坎貝爾圖
座標軸: 自然頻率與轉速。
顯示: 臨界轉速 將 發生位置(預測性)。基於特徵值分析或從瀑布圖資料中提取。
波德圖
座標軸: vibration amplitude & phase vs. rotational speed.
顯示: 實際升速/惰轉降速期間的測量響應。確認臨界轉速位置並提供用於裕度計算的放大係數。
瀑布圖(級聯圖)
座標軸: 頻譜與轉速(3D)。
顯示: 每個 RPM 步進下的完整頻譜內容。提取實驗坎貝爾圖的原始資料。同時揭示所有激振階次。
無阻尼臨界轉速圖
座標軸: 自然頻率與軸承剛性(非轉速)。
顯示: 臨界轉速如何隨支撐剛性變化而偏移。用於早期設計,在生成完整坎貝爾圖之前確定軸承剛性範圍。
軌跡圖
座標軸: 單一轉速下的 X 位移與 Y 位移。
顯示: 特定 RPM 下軸運動的形狀。前向渦動產生圓形軌跡;後向渦動產生逆行橢圓。
穩定性圖
座標軸: 對數減縮量(或實特徵值)與轉速。
顯示: 系統穩定(正阻尼)與不穩定(負阻尼)的區域。坎貝爾圖擴展一個維度。
實務案例:高速壓縮機
考慮一台設計用於 15,000 RPM 連續運轉(250 Hz)的離心壓縮機,跳脫轉速為 17,250 RPM(115%)。
坎貝爾圖結果
- 1st FW Critical (1×): 5,200 RPM (86.7 Hz) — safely below operating range.
- 2nd FW Critical (1×): 19,800 RPM (330 Hz) — above trip speed.
- 1st FW × 2×: 2,600 RPM — only relevant during startup; passed through quickly.
Margin Check
Minimum operating speed: 12,000 RPM. Separation from 1st FW critical at 5,200 RPM:
The AF at this critical from the Bode plot is 4.2, yielding a required SM of 10.7% per the API 617 formula. Actual SM of 56.7% far exceeds the requirement — no issue.
Separation from 2nd FW critical at 19,800 RPM to trip speed 17,250 RPM:
The AF at this critical is 6.5, yielding a required SM of 13.6%. Actual SM of 14.8% passes, but marginally. The engineer flags this in the report and recommends verifying the exact AF during shop mechanical running tests.
If fouling increases impeller mass by 3%, the 2nd FW critical drops from 19,800 to approximately 19,200 RPM, reducing the separation margin to 11.3% — below the required 13.6%. This scenario must be captured in the sensitivity analysis submitted with the API datasheet.
Software Tools for Campbell Diagrams
Campbell diagrams are produced by both general-purpose FEA platforms and dedicated rotordynamics packages.
| 工具 | 類型 | 備註 |
|---|---|---|
| ANSYS Mechanical (Rotordynamics) | General FEA | Full 3D solid + beam models; built-in Campbell chart post-processor; requires damped modal analysis with RGYRO |
| Siemens Simcenter 3D | General FEA | Superelement reduction for multi-rotor systems; integrated orbit and stability plots |
| DyRoBeS | Dedicated rotordynamics | Beam-element based; fast; widely used in compressor and turbine OEMs per API 684 tutorial |
| XLTRC² (Texas A&M) | Dedicated rotordynamics | Spreadsheet-based workflow; strong bearing coefficient library; popular in pump and compressor analysis |
| MADYN 2000 | Dedicated rotordynamics | German-developed; FE + transfer-matrix hybrid; excellent for torsional + lateral coupled analyses |
| COMSOL Multiphysics | General FEA | Rotordynamics module for custom models; programmable post-processing |
| Bently Nevada System 1 / ADRE | 狀態監測 | Extracts experimental Campbell diagrams from field vibration data; real-time tracking |
Common Mistakes When Using Campbell Diagrams
1. Ignoring Gyroscopic Effects
Running an undamped, zero-speed modal analysis and assuming those frequencies are the critical speeds. This produces flat lines that miss forward/backward splitting entirely. Always solve the speed-dependent eigenvalue problem.
2. Using Too Coarse a Speed Increment
If the RPM step is 2,000 RPM in a machine running at 10,000, you might miss a narrow crossing entirely. Use increments of 100–500 RPM for reliable curve definition.
3. Confusing Campbell and Bode
The Campbell diagram predicts 其中 criticals are; the Bode plot shows how severe they are. Both are required for a complete rotordynamic evaluation per API 617.
4. Neglecting Foundation and Support Flexibility
A rotor model with rigid supports will produce different critical speeds than the same rotor on a real flexible foundation. Include pedestal and foundation compliance in the model.
5. Forgetting Temperature and Load Effects
Bearing clearances change with temperature, altering stiffness coefficients. Process-gas density affects seal cross-coupling. The Campbell diagram should be run at both minimum and maximum clearance / density conditions.
6. Treating All Intersections as Equally Dangerous
A 1× intersection with the first forward mode is far more dangerous than a 4× intersection with a high backward mode. Prioritize by excitation energy and mode type.
Need On-Site Vibration Data?
Balanset-1A captures vibration spectra during run-up/coastdown for waterfall plots and experimental Campbell diagrams. Two-channel, two-plane, ISO 1940 compliant. Ships worldwide via DHL Express.
常見問題
坎貝爾圖與波德圖有何差異?
A Campbell diagram plots the system's natural frequencies against rotational speed — it predicts at which speeds critical conditions exist. A Bode plot plots actual measured (or calculated) vibration amplitude and phase against rotational speed — it shows 振幅多大 the rotor vibrates at those critical speeds. Engineers use the Campbell diagram for design and the Bode plot for verification. Both are required by API 617 for compressor certification.
API 617 要求臨界轉速的避讓裕度為何?
API 617 uses the formula SM = 17 × {1 − [1/(AF − 1.5)]}, where AF is the amplification factor at that critical speed. If AF < 2.5, no margin is required because the resonance is overdamped. For typical tilting-pad bearings (AF = 4–8), required margins range from 10% to 15%. The maximum required SM is capped at 16% for critical speeds below minimum operating speed. For critical speeds above maximum continuous speed, the same formula applies but the margin is calculated as a percentage of the maximum continuous speed.
Why do natural frequencies split into forward and backward whirl on the Campbell diagram?
Gyroscopic moments from spinning discs couple the rotor's motion in two perpendicular planes. This coupling creates two distinct precession patterns: forward whirl (precession in the same direction as shaft rotation, stiffened by the gyroscopic effect) and backward whirl (precession opposite to rotation, softened by the effect). The higher the disc's polar-to-diametral inertia ratio, the stronger the splitting. At zero speed, there is no gyroscopic moment, so both modes coalesce to a single frequency.
能否從現場量測建立坎貝爾圖?
Yes. Record vibration during a continuous startup (or coastdown) using accelerometers or proximity probes at bearing housings. Process the time-domain data into a waterfall (cascade) plot — a series of FFT spectra at each RPM increment. Extract the peak frequencies at each RPM step, then plot those peaks against RPM. The result is an experimental Campbell diagram. Coastdowns tend to give cleaner data because there are no motor-starting torque transients. Aim for a deceleration rate of 50–100 RPM/s and use at least 4,096 FFT lines for good frequency resolution.
坎貝爾圖應包含哪些激振階次?
At minimum, always include the 1× line (unbalance — the single most common excitation source in all rotating machinery). Add 2× for misalignment, shaft ovality, or cracked shafts. For turbomachinery, include blade-pass frequency (number of blades × 1×) and vane-pass frequency. For geared systems, include gear-mesh frequency. For machines with fluid-film bearings, add a 0.43–0.48× line for oil whirl. If the machine has a known defect pattern (e.g., coupling with 6 jaws), include that order (6×).
軸承類型如何影響坎貝爾圖的形狀?
Rolling-element bearings have nearly constant stiffness across the speed range, so natural-frequency curves remain almost flat (horizontal) — the only slope comes from gyroscopic effects. Fluid-film (journal) bearings increase in stiffness with speed as the oil film thins and becomes stiffer, causing natural-frequency curves to rise more steeply. Tilting-pad journal bearings behave similarly but produce less cross-coupling, improving rotor stability. Active magnetic bearings can be programmed to shift stiffness in real time, allowing engineers to reshape the Campbell diagram dynamically to avoid resonances.